Full_ML_pipeline.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00000-8b2e627e-7079-4e2c-acf3-66bcd682c2f4",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "# CEBD1260: Full Machine Learning Pipeline\n",
    "*by Hamza Benhmani & Pierre-Olivier Bonin*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00001-bd092899-03fb-4206-8e74-6e66461a28ba",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "> In this notebook, we use our custom function `utils.py` to aggregate, merge, and fill missing values of the Credit Risk datasets.\n",
    "> Please look [here](https://github.com/gzork/the-big-short/blob/main/utils.py) for the function."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "cell_id": "00002-a522186d-4530-4b7e-82b1-e80a86590f67",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 1737,
    "execution_start": 1616531649338,
    "source_hash": "25d81eb8"
   },
   "outputs": [],
   "source": [
    "import lightgbm as lgb\n",
    "import gc\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import glob\n",
    "import ntpath\n",
    "import operator\n",
    "import json\n",
    "import re\n",
    "from utils import *\n",
    "from sklearn import preprocessing\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import train_test_split,KFold,StratifiedKFold\n",
    "from sklearn.metrics import confusion_matrix,f1_score,classification_report\n",
    "from sklearn.metrics import roc_auc_score\n",
    "\n",
    "data_dir = \"data/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "cell_id": "00003-dee4f4ba-5272-4dc4-90ac-86c12a9344ad",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 10,
    "execution_start": 1616531651077,
    "source_hash": "46cbc01d",
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'POS_CASH_balance': {'MONTHS_BALANCE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CNT_INSTALMENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CNT_INSTALMENT_FUTURE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'SK_DPD': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'SK_DPD_DEF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'NAME_CONTRACT_STATUS_Active': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Amortizeddebt': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Approved': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Completed': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Demand': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Returnedtothestore': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Signed': ['sum']},\n",
       " 'credit_card_balance': {'MONTHS_BALANCE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_BALANCE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_CREDIT_LIMIT_ACTUAL': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_DRAWINGS_ATM_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_DRAWINGS_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_DRAWINGS_OTHER_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_DRAWINGS_POS_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_INST_MIN_REGULARITY': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_PAYMENT_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_PAYMENT_TOTAL_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_RECEIVABLE_PRINCIPAL': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_RECIVABLE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_TOTAL_RECEIVABLE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CNT_DRAWINGS_ATM_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CNT_DRAWINGS_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CNT_DRAWINGS_OTHER_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CNT_DRAWINGS_POS_CURRENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CNT_INSTALMENT_MATURE_CUM': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'SK_DPD': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'SK_DPD_DEF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'NAME_CONTRACT_STATUS_Active': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Completed': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Demand': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Signed': ['sum']},\n",
       " 'installments_payments': {'NUM_INSTALMENT_VERSION': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'NUM_INSTALMENT_NUMBER': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_INSTALMENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_ENTRY_PAYMENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_INSTALMENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_PAYMENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>]},\n",
       " 'previous_application': {'AMT_ANNUITY': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_APPLICATION': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_CREDIT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_DOWN_PAYMENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_GOODS_PRICE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'HOUR_APPR_PROCESS_START': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'RATE_DOWN_PAYMENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'RATE_INTEREST_PRIMARY': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'RATE_INTEREST_PRIVILEGED': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_DECISION': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'SELLERPLACE_AREA': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CNT_PAYMENT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_FIRST_DRAWING': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_FIRST_DUE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_LAST_DUE_1ST_VERSION': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_LAST_DUE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_TERMINATION': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'NAME_CONTRACT_TYPE_Cashloans': ['sum'],\n",
       "  'NAME_CONTRACT_TYPE_Consumerloans': ['sum'],\n",
       "  'NAME_CONTRACT_TYPE_Revolvingloans': ['sum'],\n",
       "  'NAME_CONTRACT_TYPE_XNA': ['sum'],\n",
       "  'WEEKDAY_APPR_PROCESS_START_FRIDAY': ['sum'],\n",
       "  'WEEKDAY_APPR_PROCESS_START_MONDAY': ['sum'],\n",
       "  'WEEKDAY_APPR_PROCESS_START_SATURDAY': ['sum'],\n",
       "  'WEEKDAY_APPR_PROCESS_START_SUNDAY': ['sum'],\n",
       "  'WEEKDAY_APPR_PROCESS_START_THURSDAY': ['sum'],\n",
       "  'WEEKDAY_APPR_PROCESS_START_TUESDAY': ['sum'],\n",
       "  'WEEKDAY_APPR_PROCESS_START_WEDNESDAY': ['sum'],\n",
       "  'FLAG_LAST_APPL_PER_CONTRACT_N': ['sum'],\n",
       "  'FLAG_LAST_APPL_PER_CONTRACT_Y': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Approved': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Canceled': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Refused': ['sum'],\n",
       "  'NAME_CONTRACT_STATUS_Unusedoffer': ['sum'],\n",
       "  'NAME_PAYMENT_TYPE_Cashthroughthebank': ['sum'],\n",
       "  'NAME_PAYMENT_TYPE_Cashlessfromtheaccountoftheemployer': ['sum'],\n",
       "  'NAME_PAYMENT_TYPE_Noncashfromyouraccount': ['sum'],\n",
       "  'NAME_PAYMENT_TYPE_XNA': ['sum'],\n",
       "  'CODE_REJECT_REASON_CLIENT': ['sum'],\n",
       "  'CODE_REJECT_REASON_HC': ['sum'],\n",
       "  'CODE_REJECT_REASON_LIMIT': ['sum'],\n",
       "  'CODE_REJECT_REASON_SCO': ['sum'],\n",
       "  'CODE_REJECT_REASON_SCOFR': ['sum'],\n",
       "  'CODE_REJECT_REASON_SYSTEM': ['sum'],\n",
       "  'CODE_REJECT_REASON_VERIF': ['sum'],\n",
       "  'CODE_REJECT_REASON_XAP': ['sum'],\n",
       "  'CODE_REJECT_REASON_XNA': ['sum'],\n",
       "  'NAME_TYPE_SUITE_Children': ['sum'],\n",
       "  'NAME_TYPE_SUITE_Family': ['sum'],\n",
       "  'NAME_TYPE_SUITE_Groupofpeople': ['sum'],\n",
       "  'NAME_TYPE_SUITE_Other_A': ['sum'],\n",
       "  'NAME_TYPE_SUITE_Other_B': ['sum'],\n",
       "  'NAME_TYPE_SUITE_Spousepartner': ['sum'],\n",
       "  'NAME_TYPE_SUITE_Unaccompanied': ['sum'],\n",
       "  'NAME_CLIENT_TYPE_New': ['sum'],\n",
       "  'NAME_CLIENT_TYPE_Refreshed': ['sum'],\n",
       "  'NAME_CLIENT_TYPE_Repeater': ['sum'],\n",
       "  'NAME_CLIENT_TYPE_XNA': ['sum'],\n",
       "  'NAME_PORTFOLIO_Cards': ['sum'],\n",
       "  'NAME_PORTFOLIO_Cars': ['sum'],\n",
       "  'NAME_PORTFOLIO_Cash': ['sum'],\n",
       "  'NAME_PORTFOLIO_POS': ['sum'],\n",
       "  'NAME_PORTFOLIO_XNA': ['sum'],\n",
       "  'NAME_PRODUCT_TYPE_XNA': ['sum'],\n",
       "  'NAME_PRODUCT_TYPE_walkin': ['sum'],\n",
       "  'NAME_PRODUCT_TYPE_xsell': ['sum'],\n",
       "  'CHANNEL_TYPE_APCashloan': ['sum'],\n",
       "  'CHANNEL_TYPE_Cardealer': ['sum'],\n",
       "  'CHANNEL_TYPE_Channelofcorporatesales': ['sum'],\n",
       "  'CHANNEL_TYPE_Contactcenter': ['sum'],\n",
       "  'CHANNEL_TYPE_Countrywide': ['sum'],\n",
       "  'CHANNEL_TYPE_Creditandcashoffices': ['sum'],\n",
       "  'CHANNEL_TYPE_RegionalLocal': ['sum'],\n",
       "  'CHANNEL_TYPE_Stone': ['sum'],\n",
       "  'NAME_YIELD_GROUP_XNA': ['sum'],\n",
       "  'NAME_YIELD_GROUP_high': ['sum'],\n",
       "  'NAME_YIELD_GROUP_low_action': ['sum'],\n",
       "  'NAME_YIELD_GROUP_low_normal': ['sum'],\n",
       "  'NAME_YIELD_GROUP_middle': ['sum']},\n",
       " 'bureau': {'SK_ID_BUREAU': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_CREDIT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CREDIT_DAY_OVERDUE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_CREDIT_ENDDATE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_ENDDATE_FACT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_CREDIT_MAX_OVERDUE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CNT_CREDIT_PROLONG': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_CREDIT_SUM': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_CREDIT_SUM_DEBT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_CREDIT_SUM_LIMIT': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_CREDIT_SUM_OVERDUE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'DAYS_CREDIT_UPDATE': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'AMT_ANNUITY': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF': ['max',\n",
       "   'min',\n",
       "   'median',\n",
       "   <function utils.aggr_dicts_json_to_dict.<locals>.<lambda>(x)>],\n",
       "  'CREDIT_ACTIVE_Active': ['sum'],\n",
       "  'CREDIT_ACTIVE_Closed': ['sum'],\n",
       "  'CREDIT_ACTIVE_Sold': ['sum'],\n",
       "  'CREDIT_CURRENCY_currency1': ['sum'],\n",
       "  'CREDIT_CURRENCY_currency2': ['sum'],\n",
       "  'CREDIT_CURRENCY_currency3': ['sum']}}"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "with open(\"aggr_dicts.json\") as json_file:\n",
    "    data_from_json = json.load(json_file)\n",
    "aggr_dicts = aggr_dicts_json_to_dict(data_from_json)\n",
    "aggr_dicts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "cell_id": "00003-891ef601-609e-4a0e-8d0a-16d655818c89",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 35133,
    "execution_start": 1616531651089,
    "source_hash": "5db598fe"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 22.7 s, sys: 3.33 s, total: 26 s\n",
      "Wall time: 27.6 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(307511, 122)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "names = [re.sub('\\.csv$', '', ntpath.basename(p)) for p in glob.glob(data_dir + \"*.csv\")]\n",
    "\n",
    "dfs = {}\n",
    "for df_name in names:\n",
    "    dfs[df_name] = pd.read_csv(f\"{data_dir}{df_name}.csv\")\n",
    "\n",
    "dfs[\"application_train\"].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "cell_id": "00004-ad509f1c-4e17-43e7-b6d0-ad651ee3615d",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 1,
    "execution_start": 1616531686215,
    "source_hash": "e1fea00e"
   },
   "outputs": [],
   "source": [
    "def train_model():\n",
    "    # main_df = preprocessing_data() we remove this since we've already done the preprocessing\n",
    "    # building model\n",
    "    model_lgb = lgb.LGBMClassifier(\n",
    "        n_jobs=6,\n",
    "        n_estimators=100000,\n",
    "        boost_from_average='false',\n",
    "        learning_rate=0.01,\n",
    "        num_leaves=64,\n",
    "        num_threads=4,\n",
    "        max_depth=-1,\n",
    "        tree_learner=\"serial\",\n",
    "        feature_fraction=0.7,\n",
    "        bagging_freq=5,\n",
    "        bagging_fraction=0.7,\n",
    "        min_data_in_leaf=600,\n",
    "        silent=-1,\n",
    "        verbose=-1,\n",
    "        max_bin=255,\n",
    "        bagging_seed=11,\n",
    "    )\n",
    "\n",
    "    features = [f for f in main_df.columns if f not in ['SK_ID_CURR', 'TARGET']]\n",
    "    X = main_df[features]\n",
    "    y = main_df['TARGET']\n",
    "\n",
    "    X_train, X_valid, y_train, y_valid = train_test_split(X, y, shuffle=True, random_state=10, test_size=0.1)\n",
    "\n",
    "    model_lgb.fit(X_train, y_train, eval_set=[(X_train, y_train), (X_valid, y_valid)], eval_metric='auc', verbose=200,\n",
    "                  early_stopping_rounds=300)\n",
    "    y_valid_lgb = model_lgb.predict_proba(X_valid, num_iteration=model_lgb.best_iteration_)[:, 1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00005-37dcce90-852b-4aa3-aa06-16e0907cb674",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "## Benchmark Model: performance is 76.3% correct prediction on 1s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "cell_id": "00006-8e996b73-5f05-4de1-bec7-a4d6632d60f7",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 105980,
    "execution_start": 1616531686218,
    "source_hash": "d840e8c3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=600, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=600\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] num_threads is set=4, n_jobs=6 will be ignored. Current value: num_threads=4\n",
      "Training until validation scores don't improve for 300 rounds\n",
      "[200]\ttraining's auc: 0.757976\ttraining's binary_logloss: 0.272726\tvalid_1's auc: 0.74671\tvalid_1's binary_logloss: 0.275005\n",
      "[400]\ttraining's auc: 0.770276\ttraining's binary_logloss: 0.245325\tvalid_1's auc: 0.75287\tvalid_1's binary_logloss: 0.250409\n",
      "[600]\ttraining's auc: 0.784534\ttraining's binary_logloss: 0.239005\tvalid_1's auc: 0.7594\tvalid_1's binary_logloss: 0.246986\n",
      "[800]\ttraining's auc: 0.797108\ttraining's binary_logloss: 0.234751\tvalid_1's auc: 0.762546\tvalid_1's binary_logloss: 0.245857\n",
      "[1000]\ttraining's auc: 0.808086\ttraining's binary_logloss: 0.231171\tvalid_1's auc: 0.76404\tvalid_1's binary_logloss: 0.245343\n",
      "[1200]\ttraining's auc: 0.818208\ttraining's binary_logloss: 0.227928\tvalid_1's auc: 0.765117\tvalid_1's binary_logloss: 0.245005\n",
      "[1400]\ttraining's auc: 0.827321\ttraining's binary_logloss: 0.22493\tvalid_1's auc: 0.765206\tvalid_1's binary_logloss: 0.24494\n",
      "[1600]\ttraining's auc: 0.835952\ttraining's binary_logloss: 0.222083\tvalid_1's auc: 0.765149\tvalid_1's binary_logloss: 0.244903\n",
      "[1800]\ttraining's auc: 0.843862\ttraining's binary_logloss: 0.219363\tvalid_1's auc: 0.765235\tvalid_1's binary_logloss: 0.244843\n",
      "Early stopping, best iteration is:\n",
      "[1560]\ttraining's auc: 0.834241\ttraining's binary_logloss: 0.222639\tvalid_1's auc: 0.765379\tvalid_1's binary_logloss: 0.244857\n"
     ]
    }
   ],
   "source": [
    "main_df = dfs['application_train']\n",
    "na_catfiller(main_df)\n",
    "na_numfiller(main_df)\n",
    "main_df = str_catencoder(main_df)\n",
    "\n",
    "train_model()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00007-cfa65c40-6ad8-4d9e-a3be-887bd9179012",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "## Aggregations and Merging"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "cell_id": "00008-afeb92dd-4bf7-4ee4-86b5-3ec43a955334",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 147020,
    "execution_start": 1616531792199,
    "source_hash": "ee30754e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 51.9 s, sys: 2.25 s, total: 54.1 s\n",
      "Wall time: 55.7 s\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SK_ID_BUREAU</th>\n",
       "      <th>STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF</th>\n",
       "      <th>STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF</th>\n",
       "      <th>STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF</th>\n",
       "      <th>STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF</th>\n",
       "      <th>STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF</th>\n",
       "      <th>STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF</th>\n",
       "      <th>STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF</th>\n",
       "      <th>STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5001709</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>85.0</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5001710</td>\n",
       "      <td>5.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>47.0</td>\n",
       "      <td>33.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5001711</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5001712</td>\n",
       "      <td>9.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>8.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5001713</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>21.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>817390</th>\n",
       "      <td>6842884</td>\n",
       "      <td>8.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>19.0</td>\n",
       "      <td>27.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>817391</th>\n",
       "      <td>6842885</td>\n",
       "      <td>11.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>11.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>817392</th>\n",
       "      <td>6842886</td>\n",
       "      <td>7.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>24.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>817393</th>\n",
       "      <td>6842887</td>\n",
       "      <td>5.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>30.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>817394</th>\n",
       "      <td>6842888</td>\n",
       "      <td>9.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>51.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>817395 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        SK_ID_BUREAU  STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF  \\\n",
       "0            5001709                                   NaN   \n",
       "1            5001710                                   5.0   \n",
       "2            5001711                                   2.0   \n",
       "3            5001712                                   9.0   \n",
       "4            5001713                                   NaN   \n",
       "...              ...                                   ...   \n",
       "817390       6842884                                   8.0   \n",
       "817391       6842885                                  11.0   \n",
       "817392       6842886                                   7.0   \n",
       "817393       6842887                                   5.0   \n",
       "817394       6842888                                   9.0   \n",
       "\n",
       "        STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF  \\\n",
       "0                                        NaN   \n",
       "1                                        NaN   \n",
       "2                                        NaN   \n",
       "3                                        NaN   \n",
       "4                                        NaN   \n",
       "...                                      ...   \n",
       "817390                                   NaN   \n",
       "817391                                   NaN   \n",
       "817392                                   NaN   \n",
       "817393                                   NaN   \n",
       "817394                                   0.0   \n",
       "\n",
       "        STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF  \\\n",
       "0                                        NaN   \n",
       "1                                        NaN   \n",
       "2                                        NaN   \n",
       "3                                        NaN   \n",
       "4                                        NaN   \n",
       "...                                      ...   \n",
       "817390                                   NaN   \n",
       "817391                                   NaN   \n",
       "817392                                   NaN   \n",
       "817393                                   NaN   \n",
       "817394                                   NaN   \n",
       "\n",
       "        STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF  \\\n",
       "0                                        NaN   \n",
       "1                                        NaN   \n",
       "2                                        NaN   \n",
       "3                                        NaN   \n",
       "4                                        NaN   \n",
       "...                                      ...   \n",
       "817390                                   NaN   \n",
       "817391                                   NaN   \n",
       "817392                                   NaN   \n",
       "817393                                   NaN   \n",
       "817394                                   NaN   \n",
       "\n",
       "        STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF  \\\n",
       "0                                        NaN   \n",
       "1                                        NaN   \n",
       "2                                        NaN   \n",
       "3                                        NaN   \n",
       "4                                        NaN   \n",
       "...                                      ...   \n",
       "817390                                   NaN   \n",
       "817391                                   NaN   \n",
       "817392                                   NaN   \n",
       "817393                                   NaN   \n",
       "817394                                   NaN   \n",
       "\n",
       "        STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF  \\\n",
       "0                                        NaN   \n",
       "1                                        NaN   \n",
       "2                                        NaN   \n",
       "3                                        NaN   \n",
       "4                                        NaN   \n",
       "...                                      ...   \n",
       "817390                                   NaN   \n",
       "817391                                  11.0   \n",
       "817392                                   NaN   \n",
       "817393                                   NaN   \n",
       "817394                                   NaN   \n",
       "\n",
       "        STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF  \\\n",
       "0                                       85.0   \n",
       "1                                       47.0   \n",
       "2                                        NaN   \n",
       "3                                        8.0   \n",
       "4                                        NaN   \n",
       "...                                      ...   \n",
       "817390                                  19.0   \n",
       "817391                                   NaN   \n",
       "817392                                  24.0   \n",
       "817393                                  30.0   \n",
       "817394                                  51.0   \n",
       "\n",
       "        STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF  \n",
       "0                                       10.0  \n",
       "1                                       33.0  \n",
       "2                                        0.0  \n",
       "3                                        NaN  \n",
       "4                                       21.0  \n",
       "...                                      ...  \n",
       "817390                                  27.0  \n",
       "817391                                   NaN  \n",
       "817392                                   NaN  \n",
       "817393                                   NaN  \n",
       "817394                                   NaN  \n",
       "\n",
       "[817395 rows x 9 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "bureau_balance = dfs[\"bureau_balance\"].reset_index().groupby(['SK_ID_BUREAU', 'STATUS'])['MONTHS_BALANCE'].aggregate(lambda x: x.max()-x.min()).unstack()\n",
    "bureau_balance.columns = [f\"{bureau_balance.columns.name}_{c}_MONTHS_BALANCE_MAX_MIN_DIFF\" for c in bureau_balance.columns]\n",
    "bureau_balance.reset_index(level=0, inplace=True)\n",
    "bureau_balance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "cell_id": "00009-cff703b0-0492-49da-83dd-d8b56af5aab9",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 1564,
    "execution_start": 1616531939253,
    "source_hash": "a7c03d81"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 932 ms, sys: 287 ms, total: 1.22 s\n",
      "Wall time: 1.4 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "dfs[\"bureau\"] = dfs[\"bureau\"].merge(bureau_balance,on='SK_ID_BUREAU',how='left')\n",
    "for f in [f for f in dfs[\"bureau\"].columns if dfs[\"bureau\"][f].isnull().sum()>0]:\n",
    "    dfs[\"bureau\"][f] = dfs[\"bureau\"][f].fillna(dfs[\"bureau\"][f].median())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "for df in dfs:\n",
    "    dfs[df] = na_catfiller(dfs[df])\n",
    "    dfs[df] = str_one_hot_encoder(dfs[df], 10)\n",
    "    dfs[df] = na_numfiller(dfs[df])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "cell_id": "00010-ab3eab25-0607-4ff7-aa34-f6699e4f8a49",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 48250,
    "execution_start": 1616531940814,
    "source_hash": "d5752644"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 9min 9s, sys: 7.83 s, total: 9min 17s\n",
      "Wall time: 9min 28s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "((307511, 530),\n",
       " ['SK_ID_CURR',\n",
       "  'TARGET',\n",
       "  'CNT_CHILDREN',\n",
       "  'AMT_INCOME_TOTAL',\n",
       "  'AMT_CREDIT',\n",
       "  'AMT_ANNUITY',\n",
       "  'AMT_GOODS_PRICE',\n",
       "  'REGION_POPULATION_RELATIVE',\n",
       "  'DAYS_BIRTH',\n",
       "  'DAYS_EMPLOYED',\n",
       "  'DAYS_REGISTRATION',\n",
       "  'DAYS_ID_PUBLISH',\n",
       "  'OWN_CAR_AGE',\n",
       "  'FLAG_MOBIL',\n",
       "  'FLAG_EMP_PHONE',\n",
       "  'FLAG_WORK_PHONE',\n",
       "  'FLAG_CONT_MOBILE',\n",
       "  'FLAG_PHONE',\n",
       "  'FLAG_EMAIL',\n",
       "  'CNT_FAM_MEMBERS',\n",
       "  'REGION_RATING_CLIENT',\n",
       "  'REGION_RATING_CLIENT_W_CITY',\n",
       "  'HOUR_APPR_PROCESS_START',\n",
       "  'REG_REGION_NOT_LIVE_REGION',\n",
       "  'REG_REGION_NOT_WORK_REGION',\n",
       "  'LIVE_REGION_NOT_WORK_REGION',\n",
       "  'REG_CITY_NOT_LIVE_CITY',\n",
       "  'REG_CITY_NOT_WORK_CITY',\n",
       "  'LIVE_CITY_NOT_WORK_CITY',\n",
       "  'EXT_SOURCE_1',\n",
       "  'EXT_SOURCE_2',\n",
       "  'EXT_SOURCE_3',\n",
       "  'APARTMENTS_AVG',\n",
       "  'BASEMENTAREA_AVG',\n",
       "  'YEARS_BEGINEXPLUATATION_AVG',\n",
       "  'YEARS_BUILD_AVG',\n",
       "  'COMMONAREA_AVG',\n",
       "  'ELEVATORS_AVG',\n",
       "  'ENTRANCES_AVG',\n",
       "  'FLOORSMAX_AVG',\n",
       "  'FLOORSMIN_AVG',\n",
       "  'LANDAREA_AVG',\n",
       "  'LIVINGAPARTMENTS_AVG',\n",
       "  'LIVINGAREA_AVG',\n",
       "  'NONLIVINGAPARTMENTS_AVG',\n",
       "  'NONLIVINGAREA_AVG',\n",
       "  'APARTMENTS_MODE',\n",
       "  'BASEMENTAREA_MODE',\n",
       "  'YEARS_BEGINEXPLUATATION_MODE',\n",
       "  'YEARS_BUILD_MODE',\n",
       "  'COMMONAREA_MODE',\n",
       "  'ELEVATORS_MODE',\n",
       "  'ENTRANCES_MODE',\n",
       "  'FLOORSMAX_MODE',\n",
       "  'FLOORSMIN_MODE',\n",
       "  'LANDAREA_MODE',\n",
       "  'LIVINGAPARTMENTS_MODE',\n",
       "  'LIVINGAREA_MODE',\n",
       "  'NONLIVINGAPARTMENTS_MODE',\n",
       "  'NONLIVINGAREA_MODE',\n",
       "  'APARTMENTS_MEDI',\n",
       "  'BASEMENTAREA_MEDI',\n",
       "  'YEARS_BEGINEXPLUATATION_MEDI',\n",
       "  'YEARS_BUILD_MEDI',\n",
       "  'COMMONAREA_MEDI',\n",
       "  'ELEVATORS_MEDI',\n",
       "  'ENTRANCES_MEDI',\n",
       "  'FLOORSMAX_MEDI',\n",
       "  'FLOORSMIN_MEDI',\n",
       "  'LANDAREA_MEDI',\n",
       "  'LIVINGAPARTMENTS_MEDI',\n",
       "  'LIVINGAREA_MEDI',\n",
       "  'NONLIVINGAPARTMENTS_MEDI',\n",
       "  'NONLIVINGAREA_MEDI',\n",
       "  'TOTALAREA_MODE',\n",
       "  'OBS_30_CNT_SOCIAL_CIRCLE',\n",
       "  'DEF_30_CNT_SOCIAL_CIRCLE',\n",
       "  'OBS_60_CNT_SOCIAL_CIRCLE',\n",
       "  'DEF_60_CNT_SOCIAL_CIRCLE',\n",
       "  'DAYS_LAST_PHONE_CHANGE',\n",
       "  'FLAG_DOCUMENT_2',\n",
       "  'FLAG_DOCUMENT_3',\n",
       "  'FLAG_DOCUMENT_4',\n",
       "  'FLAG_DOCUMENT_5',\n",
       "  'FLAG_DOCUMENT_6',\n",
       "  'FLAG_DOCUMENT_7',\n",
       "  'FLAG_DOCUMENT_8',\n",
       "  'FLAG_DOCUMENT_9',\n",
       "  'FLAG_DOCUMENT_10',\n",
       "  'FLAG_DOCUMENT_11',\n",
       "  'FLAG_DOCUMENT_12',\n",
       "  'FLAG_DOCUMENT_13',\n",
       "  'FLAG_DOCUMENT_14',\n",
       "  'FLAG_DOCUMENT_15',\n",
       "  'FLAG_DOCUMENT_16',\n",
       "  'FLAG_DOCUMENT_17',\n",
       "  'FLAG_DOCUMENT_18',\n",
       "  'FLAG_DOCUMENT_19',\n",
       "  'FLAG_DOCUMENT_20',\n",
       "  'FLAG_DOCUMENT_21',\n",
       "  'AMT_REQ_CREDIT_BUREAU_HOUR',\n",
       "  'AMT_REQ_CREDIT_BUREAU_DAY',\n",
       "  'AMT_REQ_CREDIT_BUREAU_WEEK',\n",
       "  'AMT_REQ_CREDIT_BUREAU_MON',\n",
       "  'AMT_REQ_CREDIT_BUREAU_QRT',\n",
       "  'AMT_REQ_CREDIT_BUREAU_YEAR',\n",
       "  'NAME_CONTRACT_TYPE_Cashloans',\n",
       "  'NAME_CONTRACT_TYPE_Revolvingloans',\n",
       "  'CODE_GENDER_F',\n",
       "  'CODE_GENDER_M',\n",
       "  'CODE_GENDER_XNA',\n",
       "  'FLAG_OWN_CAR_N',\n",
       "  'FLAG_OWN_CAR_Y',\n",
       "  'FLAG_OWN_REALTY_N',\n",
       "  'FLAG_OWN_REALTY_Y',\n",
       "  'NAME_TYPE_SUITE_Children',\n",
       "  'NAME_TYPE_SUITE_Family',\n",
       "  'NAME_TYPE_SUITE_Groupofpeople',\n",
       "  'NAME_TYPE_SUITE_Other_A',\n",
       "  'NAME_TYPE_SUITE_Other_B',\n",
       "  'NAME_TYPE_SUITE_Spousepartner',\n",
       "  'NAME_TYPE_SUITE_Unaccompanied',\n",
       "  'NAME_TYPE_SUITE_missing_data',\n",
       "  'NAME_INCOME_TYPE_Businessman',\n",
       "  'NAME_INCOME_TYPE_Commercialassociate',\n",
       "  'NAME_INCOME_TYPE_Maternityleave',\n",
       "  'NAME_INCOME_TYPE_Pensioner',\n",
       "  'NAME_INCOME_TYPE_Stateservant',\n",
       "  'NAME_INCOME_TYPE_Student',\n",
       "  'NAME_INCOME_TYPE_Unemployed',\n",
       "  'NAME_INCOME_TYPE_Working',\n",
       "  'NAME_EDUCATION_TYPE_Academicdegree',\n",
       "  'NAME_EDUCATION_TYPE_Highereducation',\n",
       "  'NAME_EDUCATION_TYPE_Incompletehigher',\n",
       "  'NAME_EDUCATION_TYPE_Lowersecondary',\n",
       "  'NAME_EDUCATION_TYPE_Secondarysecondaryspecial',\n",
       "  'NAME_FAMILY_STATUS_Civilmarriage',\n",
       "  'NAME_FAMILY_STATUS_Married',\n",
       "  'NAME_FAMILY_STATUS_Separated',\n",
       "  'NAME_FAMILY_STATUS_Singlenotmarried',\n",
       "  'NAME_FAMILY_STATUS_Unknown',\n",
       "  'NAME_FAMILY_STATUS_Widow',\n",
       "  'NAME_HOUSING_TYPE_Coopapartment',\n",
       "  'NAME_HOUSING_TYPE_Houseapartment',\n",
       "  'NAME_HOUSING_TYPE_Municipalapartment',\n",
       "  'NAME_HOUSING_TYPE_Officeapartment',\n",
       "  'NAME_HOUSING_TYPE_Rentedapartment',\n",
       "  'NAME_HOUSING_TYPE_Withparents',\n",
       "  'WEEKDAY_APPR_PROCESS_START_FRIDAY',\n",
       "  'WEEKDAY_APPR_PROCESS_START_MONDAY',\n",
       "  'WEEKDAY_APPR_PROCESS_START_SATURDAY',\n",
       "  'WEEKDAY_APPR_PROCESS_START_SUNDAY',\n",
       "  'WEEKDAY_APPR_PROCESS_START_THURSDAY',\n",
       "  'WEEKDAY_APPR_PROCESS_START_TUESDAY',\n",
       "  'WEEKDAY_APPR_PROCESS_START_WEDNESDAY',\n",
       "  'FONDKAPREMONT_MODE_missing_data',\n",
       "  'FONDKAPREMONT_MODE_notspecified',\n",
       "  'FONDKAPREMONT_MODE_orgspecaccount',\n",
       "  'FONDKAPREMONT_MODE_regoperaccount',\n",
       "  'FONDKAPREMONT_MODE_regoperspecaccount',\n",
       "  'HOUSETYPE_MODE_blockofflats',\n",
       "  'HOUSETYPE_MODE_missing_data',\n",
       "  'HOUSETYPE_MODE_specifichousing',\n",
       "  'HOUSETYPE_MODE_terracedhouse',\n",
       "  'WALLSMATERIAL_MODE_Block',\n",
       "  'WALLSMATERIAL_MODE_Mixed',\n",
       "  'WALLSMATERIAL_MODE_Monolithic',\n",
       "  'WALLSMATERIAL_MODE_Others',\n",
       "  'WALLSMATERIAL_MODE_Panel',\n",
       "  'WALLSMATERIAL_MODE_Stonebrick',\n",
       "  'WALLSMATERIAL_MODE_Wooden',\n",
       "  'WALLSMATERIAL_MODE_missing_data',\n",
       "  'EMERGENCYSTATE_MODE_No',\n",
       "  'EMERGENCYSTATE_MODE_Yes',\n",
       "  'EMERGENCYSTATE_MODE_missing_data',\n",
       "  'POS_CASH_balance_MONTHS_BALANCE_max',\n",
       "  'POS_CASH_balance_MONTHS_BALANCE_min',\n",
       "  'POS_CASH_balance_MONTHS_BALANCE_median',\n",
       "  'POS_CASH_balance_MONTHS_BALANCE_max_min_diff',\n",
       "  'POS_CASH_balance_CNT_INSTALMENT_max',\n",
       "  'POS_CASH_balance_CNT_INSTALMENT_min',\n",
       "  'POS_CASH_balance_CNT_INSTALMENT_median',\n",
       "  'POS_CASH_balance_CNT_INSTALMENT_max_min_diff',\n",
       "  'POS_CASH_balance_CNT_INSTALMENT_FUTURE_max',\n",
       "  'POS_CASH_balance_CNT_INSTALMENT_FUTURE_min',\n",
       "  'POS_CASH_balance_CNT_INSTALMENT_FUTURE_median',\n",
       "  'POS_CASH_balance_CNT_INSTALMENT_FUTURE_max_min_diff',\n",
       "  'POS_CASH_balance_SK_DPD_max',\n",
       "  'POS_CASH_balance_SK_DPD_min',\n",
       "  'POS_CASH_balance_SK_DPD_median',\n",
       "  'POS_CASH_balance_SK_DPD_max_min_diff',\n",
       "  'POS_CASH_balance_SK_DPD_DEF_max',\n",
       "  'POS_CASH_balance_SK_DPD_DEF_min',\n",
       "  'POS_CASH_balance_SK_DPD_DEF_median',\n",
       "  'POS_CASH_balance_SK_DPD_DEF_max_min_diff',\n",
       "  'POS_CASH_balance_NAME_CONTRACT_STATUS_Active_sum',\n",
       "  'POS_CASH_balance_NAME_CONTRACT_STATUS_Amortizeddebt_sum',\n",
       "  'POS_CASH_balance_NAME_CONTRACT_STATUS_Approved_sum',\n",
       "  'POS_CASH_balance_NAME_CONTRACT_STATUS_Completed_sum',\n",
       "  'POS_CASH_balance_NAME_CONTRACT_STATUS_Demand_sum',\n",
       "  'POS_CASH_balance_NAME_CONTRACT_STATUS_Returnedtothestore_sum',\n",
       "  'POS_CASH_balance_NAME_CONTRACT_STATUS_Signed_sum',\n",
       "  'credit_card_balance_MONTHS_BALANCE_max',\n",
       "  'credit_card_balance_MONTHS_BALANCE_min',\n",
       "  'credit_card_balance_MONTHS_BALANCE_median',\n",
       "  'credit_card_balance_MONTHS_BALANCE_max_min_diff',\n",
       "  'credit_card_balance_AMT_BALANCE_max',\n",
       "  'credit_card_balance_AMT_BALANCE_min',\n",
       "  'credit_card_balance_AMT_BALANCE_median',\n",
       "  'credit_card_balance_AMT_BALANCE_max_min_diff',\n",
       "  'credit_card_balance_AMT_CREDIT_LIMIT_ACTUAL_max',\n",
       "  'credit_card_balance_AMT_CREDIT_LIMIT_ACTUAL_min',\n",
       "  'credit_card_balance_AMT_CREDIT_LIMIT_ACTUAL_median',\n",
       "  'credit_card_balance_AMT_CREDIT_LIMIT_ACTUAL_max_min_diff',\n",
       "  'credit_card_balance_AMT_DRAWINGS_ATM_CURRENT_max',\n",
       "  'credit_card_balance_AMT_DRAWINGS_ATM_CURRENT_min',\n",
       "  'credit_card_balance_AMT_DRAWINGS_ATM_CURRENT_median',\n",
       "  'credit_card_balance_AMT_DRAWINGS_ATM_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_AMT_DRAWINGS_CURRENT_max',\n",
       "  'credit_card_balance_AMT_DRAWINGS_CURRENT_min',\n",
       "  'credit_card_balance_AMT_DRAWINGS_CURRENT_median',\n",
       "  'credit_card_balance_AMT_DRAWINGS_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_AMT_DRAWINGS_OTHER_CURRENT_max',\n",
       "  'credit_card_balance_AMT_DRAWINGS_OTHER_CURRENT_min',\n",
       "  'credit_card_balance_AMT_DRAWINGS_OTHER_CURRENT_median',\n",
       "  'credit_card_balance_AMT_DRAWINGS_OTHER_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_AMT_DRAWINGS_POS_CURRENT_max',\n",
       "  'credit_card_balance_AMT_DRAWINGS_POS_CURRENT_min',\n",
       "  'credit_card_balance_AMT_DRAWINGS_POS_CURRENT_median',\n",
       "  'credit_card_balance_AMT_DRAWINGS_POS_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_AMT_INST_MIN_REGULARITY_max',\n",
       "  'credit_card_balance_AMT_INST_MIN_REGULARITY_min',\n",
       "  'credit_card_balance_AMT_INST_MIN_REGULARITY_median',\n",
       "  'credit_card_balance_AMT_INST_MIN_REGULARITY_max_min_diff',\n",
       "  'credit_card_balance_AMT_PAYMENT_CURRENT_max',\n",
       "  'credit_card_balance_AMT_PAYMENT_CURRENT_min',\n",
       "  'credit_card_balance_AMT_PAYMENT_CURRENT_median',\n",
       "  'credit_card_balance_AMT_PAYMENT_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_AMT_PAYMENT_TOTAL_CURRENT_max',\n",
       "  'credit_card_balance_AMT_PAYMENT_TOTAL_CURRENT_min',\n",
       "  'credit_card_balance_AMT_PAYMENT_TOTAL_CURRENT_median',\n",
       "  'credit_card_balance_AMT_PAYMENT_TOTAL_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_AMT_RECEIVABLE_PRINCIPAL_max',\n",
       "  'credit_card_balance_AMT_RECEIVABLE_PRINCIPAL_min',\n",
       "  'credit_card_balance_AMT_RECEIVABLE_PRINCIPAL_median',\n",
       "  'credit_card_balance_AMT_RECEIVABLE_PRINCIPAL_max_min_diff',\n",
       "  'credit_card_balance_AMT_RECIVABLE_max',\n",
       "  'credit_card_balance_AMT_RECIVABLE_min',\n",
       "  'credit_card_balance_AMT_RECIVABLE_median',\n",
       "  'credit_card_balance_AMT_RECIVABLE_max_min_diff',\n",
       "  'credit_card_balance_AMT_TOTAL_RECEIVABLE_max',\n",
       "  'credit_card_balance_AMT_TOTAL_RECEIVABLE_min',\n",
       "  'credit_card_balance_AMT_TOTAL_RECEIVABLE_median',\n",
       "  'credit_card_balance_AMT_TOTAL_RECEIVABLE_max_min_diff',\n",
       "  'credit_card_balance_CNT_DRAWINGS_ATM_CURRENT_max',\n",
       "  'credit_card_balance_CNT_DRAWINGS_ATM_CURRENT_min',\n",
       "  'credit_card_balance_CNT_DRAWINGS_ATM_CURRENT_median',\n",
       "  'credit_card_balance_CNT_DRAWINGS_ATM_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_CNT_DRAWINGS_CURRENT_max',\n",
       "  'credit_card_balance_CNT_DRAWINGS_CURRENT_min',\n",
       "  'credit_card_balance_CNT_DRAWINGS_CURRENT_median',\n",
       "  'credit_card_balance_CNT_DRAWINGS_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_CNT_DRAWINGS_OTHER_CURRENT_max',\n",
       "  'credit_card_balance_CNT_DRAWINGS_OTHER_CURRENT_min',\n",
       "  'credit_card_balance_CNT_DRAWINGS_OTHER_CURRENT_median',\n",
       "  'credit_card_balance_CNT_DRAWINGS_OTHER_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_CNT_DRAWINGS_POS_CURRENT_max',\n",
       "  'credit_card_balance_CNT_DRAWINGS_POS_CURRENT_min',\n",
       "  'credit_card_balance_CNT_DRAWINGS_POS_CURRENT_median',\n",
       "  'credit_card_balance_CNT_DRAWINGS_POS_CURRENT_max_min_diff',\n",
       "  'credit_card_balance_CNT_INSTALMENT_MATURE_CUM_max',\n",
       "  'credit_card_balance_CNT_INSTALMENT_MATURE_CUM_min',\n",
       "  'credit_card_balance_CNT_INSTALMENT_MATURE_CUM_median',\n",
       "  'credit_card_balance_CNT_INSTALMENT_MATURE_CUM_max_min_diff',\n",
       "  'credit_card_balance_SK_DPD_max',\n",
       "  'credit_card_balance_SK_DPD_min',\n",
       "  'credit_card_balance_SK_DPD_median',\n",
       "  'credit_card_balance_SK_DPD_max_min_diff',\n",
       "  'credit_card_balance_SK_DPD_DEF_max',\n",
       "  'credit_card_balance_SK_DPD_DEF_min',\n",
       "  'credit_card_balance_SK_DPD_DEF_median',\n",
       "  'credit_card_balance_SK_DPD_DEF_max_min_diff',\n",
       "  'credit_card_balance_NAME_CONTRACT_STATUS_Active_sum',\n",
       "  'credit_card_balance_NAME_CONTRACT_STATUS_Completed_sum',\n",
       "  'credit_card_balance_NAME_CONTRACT_STATUS_Demand_sum',\n",
       "  'credit_card_balance_NAME_CONTRACT_STATUS_Signed_sum',\n",
       "  'installments_payments_NUM_INSTALMENT_VERSION_max',\n",
       "  'installments_payments_NUM_INSTALMENT_VERSION_min',\n",
       "  'installments_payments_NUM_INSTALMENT_VERSION_median',\n",
       "  'installments_payments_NUM_INSTALMENT_VERSION_max_min_diff',\n",
       "  'installments_payments_NUM_INSTALMENT_NUMBER_max',\n",
       "  'installments_payments_NUM_INSTALMENT_NUMBER_min',\n",
       "  'installments_payments_NUM_INSTALMENT_NUMBER_median',\n",
       "  'installments_payments_NUM_INSTALMENT_NUMBER_max_min_diff',\n",
       "  'installments_payments_DAYS_INSTALMENT_max',\n",
       "  'installments_payments_DAYS_INSTALMENT_min',\n",
       "  'installments_payments_DAYS_INSTALMENT_median',\n",
       "  'installments_payments_DAYS_INSTALMENT_max_min_diff',\n",
       "  'installments_payments_DAYS_ENTRY_PAYMENT_max',\n",
       "  'installments_payments_DAYS_ENTRY_PAYMENT_min',\n",
       "  'installments_payments_DAYS_ENTRY_PAYMENT_median',\n",
       "  'installments_payments_DAYS_ENTRY_PAYMENT_max_min_diff',\n",
       "  'installments_payments_AMT_INSTALMENT_max',\n",
       "  'installments_payments_AMT_INSTALMENT_min',\n",
       "  'installments_payments_AMT_INSTALMENT_median',\n",
       "  'installments_payments_AMT_INSTALMENT_max_min_diff',\n",
       "  'installments_payments_AMT_PAYMENT_max',\n",
       "  'installments_payments_AMT_PAYMENT_min',\n",
       "  'installments_payments_AMT_PAYMENT_median',\n",
       "  'installments_payments_AMT_PAYMENT_max_min_diff',\n",
       "  'previous_application_AMT_ANNUITY_max',\n",
       "  'previous_application_AMT_ANNUITY_min',\n",
       "  'previous_application_AMT_ANNUITY_median',\n",
       "  'previous_application_AMT_ANNUITY_max_min_diff',\n",
       "  'previous_application_AMT_APPLICATION_max',\n",
       "  'previous_application_AMT_APPLICATION_min',\n",
       "  'previous_application_AMT_APPLICATION_median',\n",
       "  'previous_application_AMT_APPLICATION_max_min_diff',\n",
       "  'previous_application_AMT_CREDIT_max',\n",
       "  'previous_application_AMT_CREDIT_min',\n",
       "  'previous_application_AMT_CREDIT_median',\n",
       "  'previous_application_AMT_CREDIT_max_min_diff',\n",
       "  'previous_application_AMT_DOWN_PAYMENT_max',\n",
       "  'previous_application_AMT_DOWN_PAYMENT_min',\n",
       "  'previous_application_AMT_DOWN_PAYMENT_median',\n",
       "  'previous_application_AMT_DOWN_PAYMENT_max_min_diff',\n",
       "  'previous_application_AMT_GOODS_PRICE_max',\n",
       "  'previous_application_AMT_GOODS_PRICE_min',\n",
       "  'previous_application_AMT_GOODS_PRICE_median',\n",
       "  'previous_application_AMT_GOODS_PRICE_max_min_diff',\n",
       "  'previous_application_HOUR_APPR_PROCESS_START_max',\n",
       "  'previous_application_HOUR_APPR_PROCESS_START_min',\n",
       "  'previous_application_HOUR_APPR_PROCESS_START_median',\n",
       "  'previous_application_HOUR_APPR_PROCESS_START_max_min_diff',\n",
       "  'previous_application_RATE_DOWN_PAYMENT_max',\n",
       "  'previous_application_RATE_DOWN_PAYMENT_min',\n",
       "  'previous_application_RATE_DOWN_PAYMENT_median',\n",
       "  'previous_application_RATE_DOWN_PAYMENT_max_min_diff',\n",
       "  'previous_application_RATE_INTEREST_PRIMARY_max',\n",
       "  'previous_application_RATE_INTEREST_PRIMARY_min',\n",
       "  'previous_application_RATE_INTEREST_PRIMARY_median',\n",
       "  'previous_application_RATE_INTEREST_PRIMARY_max_min_diff',\n",
       "  'previous_application_RATE_INTEREST_PRIVILEGED_max',\n",
       "  'previous_application_RATE_INTEREST_PRIVILEGED_min',\n",
       "  'previous_application_RATE_INTEREST_PRIVILEGED_median',\n",
       "  'previous_application_RATE_INTEREST_PRIVILEGED_max_min_diff',\n",
       "  'previous_application_DAYS_DECISION_max',\n",
       "  'previous_application_DAYS_DECISION_min',\n",
       "  'previous_application_DAYS_DECISION_median',\n",
       "  'previous_application_DAYS_DECISION_max_min_diff',\n",
       "  'previous_application_SELLERPLACE_AREA_max',\n",
       "  'previous_application_SELLERPLACE_AREA_min',\n",
       "  'previous_application_SELLERPLACE_AREA_median',\n",
       "  'previous_application_SELLERPLACE_AREA_max_min_diff',\n",
       "  'previous_application_CNT_PAYMENT_max',\n",
       "  'previous_application_CNT_PAYMENT_min',\n",
       "  'previous_application_CNT_PAYMENT_median',\n",
       "  'previous_application_CNT_PAYMENT_max_min_diff',\n",
       "  'previous_application_DAYS_FIRST_DRAWING_max',\n",
       "  'previous_application_DAYS_FIRST_DRAWING_min',\n",
       "  'previous_application_DAYS_FIRST_DRAWING_median',\n",
       "  'previous_application_DAYS_FIRST_DRAWING_max_min_diff',\n",
       "  'previous_application_DAYS_FIRST_DUE_max',\n",
       "  'previous_application_DAYS_FIRST_DUE_min',\n",
       "  'previous_application_DAYS_FIRST_DUE_median',\n",
       "  'previous_application_DAYS_FIRST_DUE_max_min_diff',\n",
       "  'previous_application_DAYS_LAST_DUE_1ST_VERSION_max',\n",
       "  'previous_application_DAYS_LAST_DUE_1ST_VERSION_min',\n",
       "  'previous_application_DAYS_LAST_DUE_1ST_VERSION_median',\n",
       "  'previous_application_DAYS_LAST_DUE_1ST_VERSION_max_min_diff',\n",
       "  'previous_application_DAYS_LAST_DUE_max',\n",
       "  'previous_application_DAYS_LAST_DUE_min',\n",
       "  'previous_application_DAYS_LAST_DUE_median',\n",
       "  'previous_application_DAYS_LAST_DUE_max_min_diff',\n",
       "  'previous_application_DAYS_TERMINATION_max',\n",
       "  'previous_application_DAYS_TERMINATION_min',\n",
       "  'previous_application_DAYS_TERMINATION_median',\n",
       "  'previous_application_DAYS_TERMINATION_max_min_diff',\n",
       "  'previous_application_NAME_CONTRACT_TYPE_Cashloans_sum',\n",
       "  'previous_application_NAME_CONTRACT_TYPE_Consumerloans_sum',\n",
       "  'previous_application_NAME_CONTRACT_TYPE_Revolvingloans_sum',\n",
       "  'previous_application_NAME_CONTRACT_TYPE_XNA_sum',\n",
       "  'previous_application_WEEKDAY_APPR_PROCESS_START_FRIDAY_sum',\n",
       "  'previous_application_WEEKDAY_APPR_PROCESS_START_MONDAY_sum',\n",
       "  'previous_application_WEEKDAY_APPR_PROCESS_START_SATURDAY_sum',\n",
       "  'previous_application_WEEKDAY_APPR_PROCESS_START_SUNDAY_sum',\n",
       "  'previous_application_WEEKDAY_APPR_PROCESS_START_THURSDAY_sum',\n",
       "  'previous_application_WEEKDAY_APPR_PROCESS_START_TUESDAY_sum',\n",
       "  'previous_application_WEEKDAY_APPR_PROCESS_START_WEDNESDAY_sum',\n",
       "  'previous_application_FLAG_LAST_APPL_PER_CONTRACT_N_sum',\n",
       "  'previous_application_FLAG_LAST_APPL_PER_CONTRACT_Y_sum',\n",
       "  'previous_application_NAME_CONTRACT_STATUS_Approved_sum',\n",
       "  'previous_application_NAME_CONTRACT_STATUS_Canceled_sum',\n",
       "  'previous_application_NAME_CONTRACT_STATUS_Refused_sum',\n",
       "  'previous_application_NAME_CONTRACT_STATUS_Unusedoffer_sum',\n",
       "  'previous_application_NAME_PAYMENT_TYPE_Cashthroughthebank_sum',\n",
       "  'previous_application_NAME_PAYMENT_TYPE_Cashlessfromtheaccountoftheemployer_sum',\n",
       "  'previous_application_NAME_PAYMENT_TYPE_Noncashfromyouraccount_sum',\n",
       "  'previous_application_NAME_PAYMENT_TYPE_XNA_sum',\n",
       "  'previous_application_CODE_REJECT_REASON_CLIENT_sum',\n",
       "  'previous_application_CODE_REJECT_REASON_HC_sum',\n",
       "  'previous_application_CODE_REJECT_REASON_LIMIT_sum',\n",
       "  'previous_application_CODE_REJECT_REASON_SCO_sum',\n",
       "  'previous_application_CODE_REJECT_REASON_SCOFR_sum',\n",
       "  'previous_application_CODE_REJECT_REASON_SYSTEM_sum',\n",
       "  'previous_application_CODE_REJECT_REASON_VERIF_sum',\n",
       "  'previous_application_CODE_REJECT_REASON_XAP_sum',\n",
       "  'previous_application_CODE_REJECT_REASON_XNA_sum',\n",
       "  'previous_application_NAME_TYPE_SUITE_Children_sum',\n",
       "  'previous_application_NAME_TYPE_SUITE_Family_sum',\n",
       "  'previous_application_NAME_TYPE_SUITE_Groupofpeople_sum',\n",
       "  'previous_application_NAME_TYPE_SUITE_Other_A_sum',\n",
       "  'previous_application_NAME_TYPE_SUITE_Other_B_sum',\n",
       "  'previous_application_NAME_TYPE_SUITE_Spousepartner_sum',\n",
       "  'previous_application_NAME_TYPE_SUITE_Unaccompanied_sum',\n",
       "  'previous_application_NAME_CLIENT_TYPE_New_sum',\n",
       "  'previous_application_NAME_CLIENT_TYPE_Refreshed_sum',\n",
       "  'previous_application_NAME_CLIENT_TYPE_Repeater_sum',\n",
       "  'previous_application_NAME_CLIENT_TYPE_XNA_sum',\n",
       "  'previous_application_NAME_PORTFOLIO_Cards_sum',\n",
       "  'previous_application_NAME_PORTFOLIO_Cars_sum',\n",
       "  'previous_application_NAME_PORTFOLIO_Cash_sum',\n",
       "  'previous_application_NAME_PORTFOLIO_POS_sum',\n",
       "  'previous_application_NAME_PORTFOLIO_XNA_sum',\n",
       "  'previous_application_NAME_PRODUCT_TYPE_XNA_sum',\n",
       "  'previous_application_NAME_PRODUCT_TYPE_walkin_sum',\n",
       "  'previous_application_NAME_PRODUCT_TYPE_xsell_sum',\n",
       "  'previous_application_CHANNEL_TYPE_APCashloan_sum',\n",
       "  'previous_application_CHANNEL_TYPE_Cardealer_sum',\n",
       "  'previous_application_CHANNEL_TYPE_Channelofcorporatesales_sum',\n",
       "  'previous_application_CHANNEL_TYPE_Contactcenter_sum',\n",
       "  'previous_application_CHANNEL_TYPE_Countrywide_sum',\n",
       "  'previous_application_CHANNEL_TYPE_Creditandcashoffices_sum',\n",
       "  'previous_application_CHANNEL_TYPE_RegionalLocal_sum',\n",
       "  'previous_application_CHANNEL_TYPE_Stone_sum',\n",
       "  'previous_application_NAME_YIELD_GROUP_XNA_sum',\n",
       "  'previous_application_NAME_YIELD_GROUP_high_sum',\n",
       "  'previous_application_NAME_YIELD_GROUP_low_action_sum',\n",
       "  'previous_application_NAME_YIELD_GROUP_low_normal_sum',\n",
       "  'previous_application_NAME_YIELD_GROUP_middle_sum',\n",
       "  'bureau_SK_ID_BUREAU_max',\n",
       "  'bureau_SK_ID_BUREAU_min',\n",
       "  'bureau_SK_ID_BUREAU_median',\n",
       "  'bureau_SK_ID_BUREAU_max_min_diff',\n",
       "  'bureau_DAYS_CREDIT_max',\n",
       "  'bureau_DAYS_CREDIT_min',\n",
       "  'bureau_DAYS_CREDIT_median',\n",
       "  'bureau_DAYS_CREDIT_max_min_diff',\n",
       "  'bureau_CREDIT_DAY_OVERDUE_max',\n",
       "  'bureau_CREDIT_DAY_OVERDUE_min',\n",
       "  'bureau_CREDIT_DAY_OVERDUE_median',\n",
       "  'bureau_CREDIT_DAY_OVERDUE_max_min_diff',\n",
       "  'bureau_DAYS_CREDIT_ENDDATE_max',\n",
       "  'bureau_DAYS_CREDIT_ENDDATE_min',\n",
       "  'bureau_DAYS_CREDIT_ENDDATE_median',\n",
       "  'bureau_DAYS_CREDIT_ENDDATE_max_min_diff',\n",
       "  'bureau_DAYS_ENDDATE_FACT_max',\n",
       "  'bureau_DAYS_ENDDATE_FACT_min',\n",
       "  'bureau_DAYS_ENDDATE_FACT_median',\n",
       "  'bureau_DAYS_ENDDATE_FACT_max_min_diff',\n",
       "  'bureau_AMT_CREDIT_MAX_OVERDUE_max',\n",
       "  'bureau_AMT_CREDIT_MAX_OVERDUE_min',\n",
       "  'bureau_AMT_CREDIT_MAX_OVERDUE_median',\n",
       "  'bureau_AMT_CREDIT_MAX_OVERDUE_max_min_diff',\n",
       "  'bureau_CNT_CREDIT_PROLONG_max',\n",
       "  'bureau_CNT_CREDIT_PROLONG_min',\n",
       "  'bureau_CNT_CREDIT_PROLONG_median',\n",
       "  'bureau_CNT_CREDIT_PROLONG_max_min_diff',\n",
       "  'bureau_AMT_CREDIT_SUM_max',\n",
       "  'bureau_AMT_CREDIT_SUM_min',\n",
       "  'bureau_AMT_CREDIT_SUM_median',\n",
       "  'bureau_AMT_CREDIT_SUM_max_min_diff',\n",
       "  'bureau_AMT_CREDIT_SUM_DEBT_max',\n",
       "  'bureau_AMT_CREDIT_SUM_DEBT_min',\n",
       "  'bureau_AMT_CREDIT_SUM_DEBT_median',\n",
       "  'bureau_AMT_CREDIT_SUM_DEBT_max_min_diff',\n",
       "  'bureau_AMT_CREDIT_SUM_LIMIT_max',\n",
       "  'bureau_AMT_CREDIT_SUM_LIMIT_min',\n",
       "  'bureau_AMT_CREDIT_SUM_LIMIT_median',\n",
       "  'bureau_AMT_CREDIT_SUM_LIMIT_max_min_diff',\n",
       "  'bureau_AMT_CREDIT_SUM_OVERDUE_max',\n",
       "  'bureau_AMT_CREDIT_SUM_OVERDUE_min',\n",
       "  'bureau_AMT_CREDIT_SUM_OVERDUE_median',\n",
       "  'bureau_AMT_CREDIT_SUM_OVERDUE_max_min_diff',\n",
       "  'bureau_DAYS_CREDIT_UPDATE_max',\n",
       "  'bureau_DAYS_CREDIT_UPDATE_min',\n",
       "  'bureau_DAYS_CREDIT_UPDATE_median',\n",
       "  'bureau_DAYS_CREDIT_UPDATE_max_min_diff',\n",
       "  'bureau_AMT_ANNUITY_max',\n",
       "  'bureau_AMT_ANNUITY_min',\n",
       "  'bureau_AMT_ANNUITY_median',\n",
       "  'bureau_AMT_ANNUITY_max_min_diff',\n",
       "  'bureau_STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "  'bureau_STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "  'bureau_STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "  'bureau_STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "  'bureau_STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "  'bureau_STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "  'bureau_STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "  'bureau_STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "  'bureau_STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "  'bureau_STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "  'bureau_STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "  'bureau_STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "  'bureau_STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "  'bureau_STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "  'bureau_STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "  'bureau_STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "  'bureau_STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "  'bureau_STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "  'bureau_STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "  'bureau_STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "  'bureau_STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "  'bureau_STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "  'bureau_STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "  'bureau_STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "  'bureau_STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "  'bureau_STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "  'bureau_STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "  'bureau_STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "  'bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "  'bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "  'bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "  'bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "  'bureau_CREDIT_ACTIVE_Active_sum',\n",
       "  'bureau_CREDIT_ACTIVE_Closed_sum',\n",
       "  'bureau_CREDIT_ACTIVE_Sold_sum',\n",
       "  'bureau_CREDIT_CURRENCY_currency1_sum',\n",
       "  'bureau_CREDIT_CURRENCY_currency2_sum',\n",
       "  'bureau_CREDIT_CURRENCY_currency3_sum'])"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "main_df = dfs[\"application_train\"]\n",
    "for table in aggr_dicts:\n",
    "    main_df = merge_with_aggr(main_df, dfs[table], \"SK_ID_CURR\", aggr_dicts[table], table)\n",
    "\n",
    "main_df.shape, main_df.columns.tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "cell_id": "00011-4146762a-913e-4319-af45-85a79b9fc8fb",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 543,
    "execution_start": 1616531989061,
    "source_hash": "b95e8f67"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SK_ID_CURR</th>\n",
       "      <th>TARGET</th>\n",
       "      <th>CNT_CHILDREN</th>\n",
       "      <th>AMT_INCOME_TOTAL</th>\n",
       "      <th>AMT_CREDIT</th>\n",
       "      <th>AMT_ANNUITY</th>\n",
       "      <th>AMT_GOODS_PRICE</th>\n",
       "      <th>REGION_POPULATION_RELATIVE</th>\n",
       "      <th>DAYS_BIRTH</th>\n",
       "      <th>DAYS_EMPLOYED</th>\n",
       "      <th>...</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Active_sum</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Closed_sum</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Sold_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency1_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency2_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency3_sum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>100002</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>202500.0</td>\n",
       "      <td>406597.5</td>\n",
       "      <td>24700.5</td>\n",
       "      <td>351000.0</td>\n",
       "      <td>0.018801</td>\n",
       "      <td>-9461</td>\n",
       "      <td>-637</td>\n",
       "      <td>...</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>100003</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>270000.0</td>\n",
       "      <td>1293502.5</td>\n",
       "      <td>35698.5</td>\n",
       "      <td>1129500.0</td>\n",
       "      <td>0.003541</td>\n",
       "      <td>-16765</td>\n",
       "      <td>-1188</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>100004</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>67500.0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>6750.0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>0.010032</td>\n",
       "      <td>-19046</td>\n",
       "      <td>-225</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>100006</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>312682.5</td>\n",
       "      <td>29686.5</td>\n",
       "      <td>297000.0</td>\n",
       "      <td>0.008019</td>\n",
       "      <td>-19005</td>\n",
       "      <td>-3039</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>100007</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>121500.0</td>\n",
       "      <td>513000.0</td>\n",
       "      <td>21865.5</td>\n",
       "      <td>513000.0</td>\n",
       "      <td>0.028663</td>\n",
       "      <td>-19932</td>\n",
       "      <td>-3038</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 530 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   SK_ID_CURR  TARGET  CNT_CHILDREN  AMT_INCOME_TOTAL  AMT_CREDIT  \\\n",
       "0      100002       1             0          202500.0    406597.5   \n",
       "1      100003       0             0          270000.0   1293502.5   \n",
       "2      100004       0             0           67500.0    135000.0   \n",
       "3      100006       0             0          135000.0    312682.5   \n",
       "4      100007       0             0          121500.0    513000.0   \n",
       "\n",
       "   AMT_ANNUITY  AMT_GOODS_PRICE  REGION_POPULATION_RELATIVE  DAYS_BIRTH  \\\n",
       "0      24700.5         351000.0                    0.018801       -9461   \n",
       "1      35698.5        1129500.0                    0.003541      -16765   \n",
       "2       6750.0         135000.0                    0.010032      -19046   \n",
       "3      29686.5         297000.0                    0.008019      -19005   \n",
       "4      21865.5         513000.0                    0.028663      -19932   \n",
       "\n",
       "   DAYS_EMPLOYED  ...  bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max  \\\n",
       "0           -637  ...                                              8.0   \n",
       "1          -1188  ...                                              6.0   \n",
       "2           -225  ...                                              6.0   \n",
       "3          -3039  ...                                              NaN   \n",
       "4          -3038  ...                                              6.0   \n",
       "\n",
       "   bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min  \\\n",
       "0                                              0.0   \n",
       "1                                              6.0   \n",
       "2                                              6.0   \n",
       "3                                              NaN   \n",
       "4                                              6.0   \n",
       "\n",
       "   bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median  \\\n",
       "0                                                7.0    \n",
       "1                                                6.0    \n",
       "2                                                6.0    \n",
       "3                                                NaN    \n",
       "4                                                6.0    \n",
       "\n",
       "   bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff  \\\n",
       "0                                                8.0          \n",
       "1                                                0.0          \n",
       "2                                                0.0          \n",
       "3                                                NaN          \n",
       "4                                                0.0          \n",
       "\n",
       "   bureau_CREDIT_ACTIVE_Active_sum  bureau_CREDIT_ACTIVE_Closed_sum  \\\n",
       "0                              2.0                              6.0   \n",
       "1                              1.0                              3.0   \n",
       "2                              0.0                              2.0   \n",
       "3                              NaN                              NaN   \n",
       "4                              0.0                              1.0   \n",
       "\n",
       "   bureau_CREDIT_ACTIVE_Sold_sum  bureau_CREDIT_CURRENCY_currency1_sum  \\\n",
       "0                            0.0                                   8.0   \n",
       "1                            0.0                                   4.0   \n",
       "2                            0.0                                   2.0   \n",
       "3                            NaN                                   NaN   \n",
       "4                            0.0                                   1.0   \n",
       "\n",
       "   bureau_CREDIT_CURRENCY_currency2_sum  bureau_CREDIT_CURRENCY_currency3_sum  \n",
       "0                                   0.0                                   0.0  \n",
       "1                                   0.0                                   0.0  \n",
       "2                                   0.0                                   0.0  \n",
       "3                                   NaN                                   NaN  \n",
       "4                                   0.0                                   0.0  \n",
       "\n",
       "[5 rows x 530 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "cell_id": "00012-65be8253-63fd-4c61-a0b1-b4db88e1a292",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 279,
    "execution_start": 1616531989594,
    "scrolled": true,
    "source_hash": "dd2abae8"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "118057.24537037036"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.isnull().sum().sum()/216"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "cell_id": "00013-c4b14d93-6d8f-4784-a967-87090f4b373e",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 4678,
    "execution_start": 1616531989869,
    "source_hash": "ec394d7c"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SK_ID_CURR</th>\n",
       "      <th>TARGET</th>\n",
       "      <th>CNT_CHILDREN</th>\n",
       "      <th>AMT_INCOME_TOTAL</th>\n",
       "      <th>AMT_CREDIT</th>\n",
       "      <th>AMT_ANNUITY</th>\n",
       "      <th>AMT_GOODS_PRICE</th>\n",
       "      <th>REGION_POPULATION_RELATIVE</th>\n",
       "      <th>DAYS_BIRTH</th>\n",
       "      <th>DAYS_EMPLOYED</th>\n",
       "      <th>...</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Active_sum</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Closed_sum</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Sold_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency1_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency2_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency3_sum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>100002</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>202500.0</td>\n",
       "      <td>406597.5</td>\n",
       "      <td>24700.5</td>\n",
       "      <td>351000.0</td>\n",
       "      <td>0.018801</td>\n",
       "      <td>-9461</td>\n",
       "      <td>-637</td>\n",
       "      <td>...</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>100003</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>270000.0</td>\n",
       "      <td>1293502.5</td>\n",
       "      <td>35698.5</td>\n",
       "      <td>1129500.0</td>\n",
       "      <td>0.003541</td>\n",
       "      <td>-16765</td>\n",
       "      <td>-1188</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>100004</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>67500.0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>6750.0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>0.010032</td>\n",
       "      <td>-19046</td>\n",
       "      <td>-225</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>100006</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>312682.5</td>\n",
       "      <td>29686.5</td>\n",
       "      <td>297000.0</td>\n",
       "      <td>0.008019</td>\n",
       "      <td>-19005</td>\n",
       "      <td>-3039</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>100007</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>121500.0</td>\n",
       "      <td>513000.0</td>\n",
       "      <td>21865.5</td>\n",
       "      <td>513000.0</td>\n",
       "      <td>0.028663</td>\n",
       "      <td>-19932</td>\n",
       "      <td>-3038</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307506</th>\n",
       "      <td>456251</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>157500.0</td>\n",
       "      <td>254700.0</td>\n",
       "      <td>27558.0</td>\n",
       "      <td>225000.0</td>\n",
       "      <td>0.032561</td>\n",
       "      <td>-9327</td>\n",
       "      <td>-236</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307507</th>\n",
       "      <td>456252</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>72000.0</td>\n",
       "      <td>269550.0</td>\n",
       "      <td>12001.5</td>\n",
       "      <td>225000.0</td>\n",
       "      <td>0.025164</td>\n",
       "      <td>-20775</td>\n",
       "      <td>365243</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307508</th>\n",
       "      <td>456253</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>153000.0</td>\n",
       "      <td>677664.0</td>\n",
       "      <td>29979.0</td>\n",
       "      <td>585000.0</td>\n",
       "      <td>0.005002</td>\n",
       "      <td>-14966</td>\n",
       "      <td>-7921</td>\n",
       "      <td>...</td>\n",
       "      <td>23.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307509</th>\n",
       "      <td>456254</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>171000.0</td>\n",
       "      <td>370107.0</td>\n",
       "      <td>20205.0</td>\n",
       "      <td>319500.0</td>\n",
       "      <td>0.005313</td>\n",
       "      <td>-11961</td>\n",
       "      <td>-4786</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307510</th>\n",
       "      <td>456255</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>157500.0</td>\n",
       "      <td>675000.0</td>\n",
       "      <td>49117.5</td>\n",
       "      <td>675000.0</td>\n",
       "      <td>0.046220</td>\n",
       "      <td>-16856</td>\n",
       "      <td>-1262</td>\n",
       "      <td>...</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>307511 rows × 530 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        SK_ID_CURR  TARGET  CNT_CHILDREN  AMT_INCOME_TOTAL  AMT_CREDIT  \\\n",
       "0           100002       1             0          202500.0    406597.5   \n",
       "1           100003       0             0          270000.0   1293502.5   \n",
       "2           100004       0             0           67500.0    135000.0   \n",
       "3           100006       0             0          135000.0    312682.5   \n",
       "4           100007       0             0          121500.0    513000.0   \n",
       "...            ...     ...           ...               ...         ...   \n",
       "307506      456251       0             0          157500.0    254700.0   \n",
       "307507      456252       0             0           72000.0    269550.0   \n",
       "307508      456253       0             0          153000.0    677664.0   \n",
       "307509      456254       1             0          171000.0    370107.0   \n",
       "307510      456255       0             0          157500.0    675000.0   \n",
       "\n",
       "        AMT_ANNUITY  AMT_GOODS_PRICE  REGION_POPULATION_RELATIVE  DAYS_BIRTH  \\\n",
       "0           24700.5         351000.0                    0.018801       -9461   \n",
       "1           35698.5        1129500.0                    0.003541      -16765   \n",
       "2            6750.0         135000.0                    0.010032      -19046   \n",
       "3           29686.5         297000.0                    0.008019      -19005   \n",
       "4           21865.5         513000.0                    0.028663      -19932   \n",
       "...             ...              ...                         ...         ...   \n",
       "307506      27558.0         225000.0                    0.032561       -9327   \n",
       "307507      12001.5         225000.0                    0.025164      -20775   \n",
       "307508      29979.0         585000.0                    0.005002      -14966   \n",
       "307509      20205.0         319500.0                    0.005313      -11961   \n",
       "307510      49117.5         675000.0                    0.046220      -16856   \n",
       "\n",
       "        DAYS_EMPLOYED  ...  bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max  \\\n",
       "0                -637  ...                                              8.0   \n",
       "1               -1188  ...                                              6.0   \n",
       "2                -225  ...                                              6.0   \n",
       "3               -3039  ...                                              6.0   \n",
       "4               -3038  ...                                              6.0   \n",
       "...               ...  ...                                              ...   \n",
       "307506           -236  ...                                              6.0   \n",
       "307507         365243  ...                                              6.0   \n",
       "307508          -7921  ...                                             23.0   \n",
       "307509          -4786  ...                                              6.0   \n",
       "307510          -1262  ...                                              8.0   \n",
       "\n",
       "        bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min  \\\n",
       "0                                                   0.0   \n",
       "1                                                   6.0   \n",
       "2                                                   6.0   \n",
       "3                                                   6.0   \n",
       "4                                                   6.0   \n",
       "...                                                 ...   \n",
       "307506                                              6.0   \n",
       "307507                                              6.0   \n",
       "307508                                              6.0   \n",
       "307509                                              6.0   \n",
       "307510                                              0.0   \n",
       "\n",
       "        bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median  \\\n",
       "0                                                     7.0    \n",
       "1                                                     6.0    \n",
       "2                                                     6.0    \n",
       "3                                                     6.0    \n",
       "4                                                     6.0    \n",
       "...                                                   ...    \n",
       "307506                                                6.0    \n",
       "307507                                                6.0    \n",
       "307508                                                6.0    \n",
       "307509                                                6.0    \n",
       "307510                                                7.0    \n",
       "\n",
       "        bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff  \\\n",
       "0                                                     8.0          \n",
       "1                                                     0.0          \n",
       "2                                                     0.0          \n",
       "3                                                     0.0          \n",
       "4                                                     0.0          \n",
       "...                                                   ...          \n",
       "307506                                                0.0          \n",
       "307507                                                0.0          \n",
       "307508                                               17.0          \n",
       "307509                                                0.0          \n",
       "307510                                                8.0          \n",
       "\n",
       "        bureau_CREDIT_ACTIVE_Active_sum  bureau_CREDIT_ACTIVE_Closed_sum  \\\n",
       "0                                   2.0                              6.0   \n",
       "1                                   1.0                              3.0   \n",
       "2                                   0.0                              2.0   \n",
       "3                                   2.0                              3.0   \n",
       "4                                   0.0                              1.0   \n",
       "...                                 ...                              ...   \n",
       "307506                              2.0                              3.0   \n",
       "307507                              2.0                              3.0   \n",
       "307508                              2.0                              2.0   \n",
       "307509                              0.0                              1.0   \n",
       "307510                              5.0                              6.0   \n",
       "\n",
       "        bureau_CREDIT_ACTIVE_Sold_sum  bureau_CREDIT_CURRENCY_currency1_sum  \\\n",
       "0                                 0.0                                   8.0   \n",
       "1                                 0.0                                   4.0   \n",
       "2                                 0.0                                   2.0   \n",
       "3                                 0.0                                   4.0   \n",
       "4                                 0.0                                   1.0   \n",
       "...                               ...                                   ...   \n",
       "307506                            0.0                                   4.0   \n",
       "307507                            0.0                                   4.0   \n",
       "307508                            0.0                                   4.0   \n",
       "307509                            0.0                                   1.0   \n",
       "307510                            0.0                                  11.0   \n",
       "\n",
       "        bureau_CREDIT_CURRENCY_currency2_sum  \\\n",
       "0                                        0.0   \n",
       "1                                        0.0   \n",
       "2                                        0.0   \n",
       "3                                        0.0   \n",
       "4                                        0.0   \n",
       "...                                      ...   \n",
       "307506                                   0.0   \n",
       "307507                                   0.0   \n",
       "307508                                   0.0   \n",
       "307509                                   0.0   \n",
       "307510                                   0.0   \n",
       "\n",
       "        bureau_CREDIT_CURRENCY_currency3_sum  \n",
       "0                                        0.0  \n",
       "1                                        0.0  \n",
       "2                                        0.0  \n",
       "3                                        0.0  \n",
       "4                                        0.0  \n",
       "...                                      ...  \n",
       "307506                                   0.0  \n",
       "307507                                   0.0  \n",
       "307508                                   0.0  \n",
       "307509                                   0.0  \n",
       "307510                                   0.0  \n",
       "\n",
       "[307511 rows x 530 columns]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "na_numfiller(main_df, aggregation_func=\"median\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "cell_id": "00014-1b3f4d40-609d-49af-a4b8-ca67526f0f06",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 281,
    "execution_start": 1616531994544,
    "source_hash": "29df4a19"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "441212"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.isnull().sum().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "cell_id": "00015-af362547-b193-4ef9-b19c-07ea06a5e7fd",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 2355,
    "execution_start": 1616531994820,
    "source_hash": "5e7201e9"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SK_ID_CURR</th>\n",
       "      <th>TARGET</th>\n",
       "      <th>CNT_CHILDREN</th>\n",
       "      <th>AMT_INCOME_TOTAL</th>\n",
       "      <th>AMT_CREDIT</th>\n",
       "      <th>AMT_ANNUITY</th>\n",
       "      <th>AMT_GOODS_PRICE</th>\n",
       "      <th>REGION_POPULATION_RELATIVE</th>\n",
       "      <th>DAYS_BIRTH</th>\n",
       "      <th>DAYS_EMPLOYED</th>\n",
       "      <th>...</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Active_sum</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Closed_sum</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Sold_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency1_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency2_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency3_sum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>100002</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>202500.0</td>\n",
       "      <td>406597.5</td>\n",
       "      <td>24700.5</td>\n",
       "      <td>351000.0</td>\n",
       "      <td>0.018801</td>\n",
       "      <td>-9461</td>\n",
       "      <td>-637</td>\n",
       "      <td>...</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>100003</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>270000.0</td>\n",
       "      <td>1293502.5</td>\n",
       "      <td>35698.5</td>\n",
       "      <td>1129500.0</td>\n",
       "      <td>0.003541</td>\n",
       "      <td>-16765</td>\n",
       "      <td>-1188</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>100004</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>67500.0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>6750.0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>0.010032</td>\n",
       "      <td>-19046</td>\n",
       "      <td>-225</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>100006</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>312682.5</td>\n",
       "      <td>29686.5</td>\n",
       "      <td>297000.0</td>\n",
       "      <td>0.008019</td>\n",
       "      <td>-19005</td>\n",
       "      <td>-3039</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>100007</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>121500.0</td>\n",
       "      <td>513000.0</td>\n",
       "      <td>21865.5</td>\n",
       "      <td>513000.0</td>\n",
       "      <td>0.028663</td>\n",
       "      <td>-19932</td>\n",
       "      <td>-3038</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307506</th>\n",
       "      <td>456251</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>157500.0</td>\n",
       "      <td>254700.0</td>\n",
       "      <td>27558.0</td>\n",
       "      <td>225000.0</td>\n",
       "      <td>0.032561</td>\n",
       "      <td>-9327</td>\n",
       "      <td>-236</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307507</th>\n",
       "      <td>456252</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>72000.0</td>\n",
       "      <td>269550.0</td>\n",
       "      <td>12001.5</td>\n",
       "      <td>225000.0</td>\n",
       "      <td>0.025164</td>\n",
       "      <td>-20775</td>\n",
       "      <td>365243</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307508</th>\n",
       "      <td>456253</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>153000.0</td>\n",
       "      <td>677664.0</td>\n",
       "      <td>29979.0</td>\n",
       "      <td>585000.0</td>\n",
       "      <td>0.005002</td>\n",
       "      <td>-14966</td>\n",
       "      <td>-7921</td>\n",
       "      <td>...</td>\n",
       "      <td>23.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307509</th>\n",
       "      <td>456254</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>171000.0</td>\n",
       "      <td>370107.0</td>\n",
       "      <td>20205.0</td>\n",
       "      <td>319500.0</td>\n",
       "      <td>0.005313</td>\n",
       "      <td>-11961</td>\n",
       "      <td>-4786</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307510</th>\n",
       "      <td>456255</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>157500.0</td>\n",
       "      <td>675000.0</td>\n",
       "      <td>49117.5</td>\n",
       "      <td>675000.0</td>\n",
       "      <td>0.046220</td>\n",
       "      <td>-16856</td>\n",
       "      <td>-1262</td>\n",
       "      <td>...</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>307511 rows × 530 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        SK_ID_CURR  TARGET  CNT_CHILDREN  AMT_INCOME_TOTAL  AMT_CREDIT  \\\n",
       "0           100002       1             0          202500.0    406597.5   \n",
       "1           100003       0             0          270000.0   1293502.5   \n",
       "2           100004       0             0           67500.0    135000.0   \n",
       "3           100006       0             0          135000.0    312682.5   \n",
       "4           100007       0             0          121500.0    513000.0   \n",
       "...            ...     ...           ...               ...         ...   \n",
       "307506      456251       0             0          157500.0    254700.0   \n",
       "307507      456252       0             0           72000.0    269550.0   \n",
       "307508      456253       0             0          153000.0    677664.0   \n",
       "307509      456254       1             0          171000.0    370107.0   \n",
       "307510      456255       0             0          157500.0    675000.0   \n",
       "\n",
       "        AMT_ANNUITY  AMT_GOODS_PRICE  REGION_POPULATION_RELATIVE  DAYS_BIRTH  \\\n",
       "0           24700.5         351000.0                    0.018801       -9461   \n",
       "1           35698.5        1129500.0                    0.003541      -16765   \n",
       "2            6750.0         135000.0                    0.010032      -19046   \n",
       "3           29686.5         297000.0                    0.008019      -19005   \n",
       "4           21865.5         513000.0                    0.028663      -19932   \n",
       "...             ...              ...                         ...         ...   \n",
       "307506      27558.0         225000.0                    0.032561       -9327   \n",
       "307507      12001.5         225000.0                    0.025164      -20775   \n",
       "307508      29979.0         585000.0                    0.005002      -14966   \n",
       "307509      20205.0         319500.0                    0.005313      -11961   \n",
       "307510      49117.5         675000.0                    0.046220      -16856   \n",
       "\n",
       "        DAYS_EMPLOYED  ...  bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max  \\\n",
       "0                -637  ...                                              8.0   \n",
       "1               -1188  ...                                              6.0   \n",
       "2                -225  ...                                              6.0   \n",
       "3               -3039  ...                                              6.0   \n",
       "4               -3038  ...                                              6.0   \n",
       "...               ...  ...                                              ...   \n",
       "307506           -236  ...                                              6.0   \n",
       "307507         365243  ...                                              6.0   \n",
       "307508          -7921  ...                                             23.0   \n",
       "307509          -4786  ...                                              6.0   \n",
       "307510          -1262  ...                                              8.0   \n",
       "\n",
       "        bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min  \\\n",
       "0                                                   0.0   \n",
       "1                                                   6.0   \n",
       "2                                                   6.0   \n",
       "3                                                   6.0   \n",
       "4                                                   6.0   \n",
       "...                                                 ...   \n",
       "307506                                              6.0   \n",
       "307507                                              6.0   \n",
       "307508                                              6.0   \n",
       "307509                                              6.0   \n",
       "307510                                              0.0   \n",
       "\n",
       "        bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median  \\\n",
       "0                                                     7.0    \n",
       "1                                                     6.0    \n",
       "2                                                     6.0    \n",
       "3                                                     6.0    \n",
       "4                                                     6.0    \n",
       "...                                                   ...    \n",
       "307506                                                6.0    \n",
       "307507                                                6.0    \n",
       "307508                                                6.0    \n",
       "307509                                                6.0    \n",
       "307510                                                7.0    \n",
       "\n",
       "        bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff  \\\n",
       "0                                                     8.0          \n",
       "1                                                     0.0          \n",
       "2                                                     0.0          \n",
       "3                                                     0.0          \n",
       "4                                                     0.0          \n",
       "...                                                   ...          \n",
       "307506                                                0.0          \n",
       "307507                                                0.0          \n",
       "307508                                               17.0          \n",
       "307509                                                0.0          \n",
       "307510                                                8.0          \n",
       "\n",
       "        bureau_CREDIT_ACTIVE_Active_sum  bureau_CREDIT_ACTIVE_Closed_sum  \\\n",
       "0                                   2.0                              6.0   \n",
       "1                                   1.0                              3.0   \n",
       "2                                   0.0                              2.0   \n",
       "3                                   2.0                              3.0   \n",
       "4                                   0.0                              1.0   \n",
       "...                                 ...                              ...   \n",
       "307506                              2.0                              3.0   \n",
       "307507                              2.0                              3.0   \n",
       "307508                              2.0                              2.0   \n",
       "307509                              0.0                              1.0   \n",
       "307510                              5.0                              6.0   \n",
       "\n",
       "        bureau_CREDIT_ACTIVE_Sold_sum  bureau_CREDIT_CURRENCY_currency1_sum  \\\n",
       "0                                 0.0                                   8.0   \n",
       "1                                 0.0                                   4.0   \n",
       "2                                 0.0                                   2.0   \n",
       "3                                 0.0                                   4.0   \n",
       "4                                 0.0                                   1.0   \n",
       "...                               ...                                   ...   \n",
       "307506                            0.0                                   4.0   \n",
       "307507                            0.0                                   4.0   \n",
       "307508                            0.0                                   4.0   \n",
       "307509                            0.0                                   1.0   \n",
       "307510                            0.0                                  11.0   \n",
       "\n",
       "        bureau_CREDIT_CURRENCY_currency2_sum  \\\n",
       "0                                        0.0   \n",
       "1                                        0.0   \n",
       "2                                        0.0   \n",
       "3                                        0.0   \n",
       "4                                        0.0   \n",
       "...                                      ...   \n",
       "307506                                   0.0   \n",
       "307507                                   0.0   \n",
       "307508                                   0.0   \n",
       "307509                                   0.0   \n",
       "307510                                   0.0   \n",
       "\n",
       "        bureau_CREDIT_CURRENCY_currency3_sum  \n",
       "0                                        0.0  \n",
       "1                                        0.0  \n",
       "2                                        0.0  \n",
       "3                                        0.0  \n",
       "4                                        0.0  \n",
       "...                                      ...  \n",
       "307506                                   0.0  \n",
       "307507                                   0.0  \n",
       "307508                                   0.0  \n",
       "307509                                   0.0  \n",
       "307510                                   0.0  \n",
       "\n",
       "[307511 rows x 530 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "na_catfiller(main_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "cell_id": "00016-ed0fdd77-c949-418f-9ce7-bc7af396fe68",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 280,
    "execution_start": 1616531997169,
    "scrolled": true,
    "source_hash": "29df4a19"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "441212"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.isnull().sum().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "cell_id": "00017-0694b4e5-a61f-4689-b243-43af7f4ffa39",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 562,
    "execution_start": 1616531997447,
    "scrolled": true,
    "source_hash": "e8235135"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Memory usage of dataframe is 1.157779 GB\n",
      "Total initial memory used for selected dataframe is: 1.16GB\n",
      "Memory usage of dataframe is 1.107347 GB\n",
      "Total memory currently used for selected dataframe is: 1.11GB\n"
     ]
    }
   ],
   "source": [
    "optimize_inttypes(main_df, specify=\"int32\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "cell_id": "00018-4026b218-e55e-4c37-8168-035495c4c7ed",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 12689,
    "execution_start": 1616531997947,
    "scrolled": true,
    "source_hash": "afcbd5d9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Memory usage of dataframe is 1.107347 GB\n",
      "Total initial memory used for selected dataframe is: 1.11GB\n",
      "Memory usage of dataframe is 0.590729 GB\n",
      "Total memory currently used for selected dataframe is: 0.59GB\n"
     ]
    }
   ],
   "source": [
    "optimize_floattypes(main_df, specify=\"float32\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "cell_id": "00019-eb600f71-ccf7-4f89-b148-d2179f4af372",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 8,
    "execution_start": 1616532010639,
    "source_hash": "b1041925"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(307511, 530)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "cell_id": "00020-3381bd73-9ef4-4af2-8ab0-31fcf03fff72",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 7117,
    "execution_start": 1616532010643,
    "scrolled": true,
    "source_hash": "dec95c99"
   },
   "outputs": [],
   "source": [
    "main_df = str_catencoder(main_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "cell_id": "00021-f97c37e9-d071-4bc8-a267-8034825f469c",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 7,
    "execution_start": 1616532017763,
    "source_hash": "b1041925"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(307511, 530)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "cell_id": "00022-4b75cd5a-c061-448c-a325-dbd50678e1b0",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 125,
    "execution_start": 1616532017766,
    "source_hash": "29df4a19"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "441212"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.isnull().sum().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "cell_id": "00023-d8bde551-9996-417c-8653-bdd8cfa2d925",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 478,
    "execution_start": 1616532017899,
    "source_hash": "b95e8f67"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SK_ID_CURR</th>\n",
       "      <th>TARGET</th>\n",
       "      <th>CNT_CHILDREN</th>\n",
       "      <th>AMT_INCOME_TOTAL</th>\n",
       "      <th>AMT_CREDIT</th>\n",
       "      <th>AMT_ANNUITY</th>\n",
       "      <th>AMT_GOODS_PRICE</th>\n",
       "      <th>REGION_POPULATION_RELATIVE</th>\n",
       "      <th>DAYS_BIRTH</th>\n",
       "      <th>DAYS_EMPLOYED</th>\n",
       "      <th>...</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median</th>\n",
       "      <th>bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Active_sum</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Closed_sum</th>\n",
       "      <th>bureau_CREDIT_ACTIVE_Sold_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency1_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency2_sum</th>\n",
       "      <th>bureau_CREDIT_CURRENCY_currency3_sum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>100002</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>202500.0</td>\n",
       "      <td>406597.5</td>\n",
       "      <td>24700.5</td>\n",
       "      <td>351000.0</td>\n",
       "      <td>0.018801</td>\n",
       "      <td>-9461</td>\n",
       "      <td>-637</td>\n",
       "      <td>...</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>100003</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>270000.0</td>\n",
       "      <td>1293502.5</td>\n",
       "      <td>35698.5</td>\n",
       "      <td>1129500.0</td>\n",
       "      <td>0.003541</td>\n",
       "      <td>-16765</td>\n",
       "      <td>-1188</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>100004</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>67500.0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>6750.0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>0.010032</td>\n",
       "      <td>-19046</td>\n",
       "      <td>-225</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>100006</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>135000.0</td>\n",
       "      <td>312682.5</td>\n",
       "      <td>29686.5</td>\n",
       "      <td>297000.0</td>\n",
       "      <td>0.008019</td>\n",
       "      <td>-19005</td>\n",
       "      <td>-3039</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>100007</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>121500.0</td>\n",
       "      <td>513000.0</td>\n",
       "      <td>21865.5</td>\n",
       "      <td>513000.0</td>\n",
       "      <td>0.028663</td>\n",
       "      <td>-19932</td>\n",
       "      <td>-3038</td>\n",
       "      <td>...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 530 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   SK_ID_CURR  TARGET  CNT_CHILDREN  AMT_INCOME_TOTAL  AMT_CREDIT  \\\n",
       "0      100002       1             0          202500.0    406597.5   \n",
       "1      100003       0             0          270000.0   1293502.5   \n",
       "2      100004       0             0           67500.0    135000.0   \n",
       "3      100006       0             0          135000.0    312682.5   \n",
       "4      100007       0             0          121500.0    513000.0   \n",
       "\n",
       "   AMT_ANNUITY  AMT_GOODS_PRICE  REGION_POPULATION_RELATIVE  DAYS_BIRTH  \\\n",
       "0      24700.5         351000.0                    0.018801       -9461   \n",
       "1      35698.5        1129500.0                    0.003541      -16765   \n",
       "2       6750.0         135000.0                    0.010032      -19046   \n",
       "3      29686.5         297000.0                    0.008019      -19005   \n",
       "4      21865.5         513000.0                    0.028663      -19932   \n",
       "\n",
       "   DAYS_EMPLOYED  ...  bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max  \\\n",
       "0           -637  ...                                              8.0   \n",
       "1          -1188  ...                                              6.0   \n",
       "2           -225  ...                                              6.0   \n",
       "3          -3039  ...                                              6.0   \n",
       "4          -3038  ...                                              6.0   \n",
       "\n",
       "   bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min  \\\n",
       "0                                              0.0   \n",
       "1                                              6.0   \n",
       "2                                              6.0   \n",
       "3                                              6.0   \n",
       "4                                              6.0   \n",
       "\n",
       "   bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median  \\\n",
       "0                                                7.0    \n",
       "1                                                6.0    \n",
       "2                                                6.0    \n",
       "3                                                6.0    \n",
       "4                                                6.0    \n",
       "\n",
       "   bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff  \\\n",
       "0                                                8.0          \n",
       "1                                                0.0          \n",
       "2                                                0.0          \n",
       "3                                                0.0          \n",
       "4                                                0.0          \n",
       "\n",
       "   bureau_CREDIT_ACTIVE_Active_sum  bureau_CREDIT_ACTIVE_Closed_sum  \\\n",
       "0                              2.0                              6.0   \n",
       "1                              1.0                              3.0   \n",
       "2                              0.0                              2.0   \n",
       "3                              2.0                              3.0   \n",
       "4                              0.0                              1.0   \n",
       "\n",
       "   bureau_CREDIT_ACTIVE_Sold_sum  bureau_CREDIT_CURRENCY_currency1_sum  \\\n",
       "0                            0.0                                   8.0   \n",
       "1                            0.0                                   4.0   \n",
       "2                            0.0                                   2.0   \n",
       "3                            0.0                                   4.0   \n",
       "4                            0.0                                   1.0   \n",
       "\n",
       "   bureau_CREDIT_CURRENCY_currency2_sum  bureau_CREDIT_CURRENCY_currency3_sum  \n",
       "0                                   0.0                                   0.0  \n",
       "1                                   0.0                                   0.0  \n",
       "2                                   0.0                                   0.0  \n",
       "3                                   0.0                                   0.0  \n",
       "4                                   0.0                                   0.0  \n",
       "\n",
       "[5 rows x 530 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00024-fd39611f-1ab7-4cf8-b056-feef06776b85",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "# Train-Test Split and Modelling using LightGBM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "cell_id": "00025-6038deab-a1f7-4b9f-b923-dcc19eea2b04",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 220926,
    "execution_start": 1616532018400,
    "source_hash": "7a162d9f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=600, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=600\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] num_threads is set=4, n_jobs=6 will be ignored. Current value: num_threads=4\n",
      "Training until validation scores don't improve for 300 rounds\n",
      "[200]\ttraining's auc: 0.772987\ttraining's binary_logloss: 0.270475\tvalid_1's auc: 0.752015\tvalid_1's binary_logloss: 0.274576\n",
      "[400]\ttraining's auc: 0.791743\ttraining's binary_logloss: 0.239449\tvalid_1's auc: 0.762402\tvalid_1's binary_logloss: 0.248022\n",
      "[600]\ttraining's auc: 0.811652\ttraining's binary_logloss: 0.230046\tvalid_1's auc: 0.772847\tvalid_1's binary_logloss: 0.243013\n",
      "[800]\ttraining's auc: 0.827252\ttraining's binary_logloss: 0.223568\tvalid_1's auc: 0.778663\tvalid_1's binary_logloss: 0.240821\n",
      "[1000]\ttraining's auc: 0.840373\ttraining's binary_logloss: 0.21823\tvalid_1's auc: 0.781565\tvalid_1's binary_logloss: 0.23973\n",
      "[1200]\ttraining's auc: 0.85259\ttraining's binary_logloss: 0.21344\tvalid_1's auc: 0.783436\tvalid_1's binary_logloss: 0.239087\n",
      "[1400]\ttraining's auc: 0.863403\ttraining's binary_logloss: 0.209076\tvalid_1's auc: 0.784382\tvalid_1's binary_logloss: 0.238702\n",
      "[1600]\ttraining's auc: 0.873327\ttraining's binary_logloss: 0.204999\tvalid_1's auc: 0.785161\tvalid_1's binary_logloss: 0.238416\n",
      "[1800]\ttraining's auc: 0.882332\ttraining's binary_logloss: 0.201067\tvalid_1's auc: 0.785328\tvalid_1's binary_logloss: 0.238322\n",
      "[2000]\ttraining's auc: 0.890976\ttraining's binary_logloss: 0.197308\tvalid_1's auc: 0.78572\tvalid_1's binary_logloss: 0.238192\n",
      "[2200]\ttraining's auc: 0.898706\ttraining's binary_logloss: 0.193702\tvalid_1's auc: 0.786002\tvalid_1's binary_logloss: 0.238094\n",
      "[2400]\ttraining's auc: 0.90596\ttraining's binary_logloss: 0.190205\tvalid_1's auc: 0.786144\tvalid_1's binary_logloss: 0.237987\n",
      "[2600]\ttraining's auc: 0.912908\ttraining's binary_logloss: 0.186764\tvalid_1's auc: 0.786333\tvalid_1's binary_logloss: 0.237941\n",
      "[2800]\ttraining's auc: 0.919228\ttraining's binary_logloss: 0.183436\tvalid_1's auc: 0.786346\tvalid_1's binary_logloss: 0.237954\n",
      "Early stopping, best iteration is:\n",
      "[2615]\ttraining's auc: 0.913419\ttraining's binary_logloss: 0.186508\tvalid_1's auc: 0.786341\tvalid_1's binary_logloss: 0.237927\n"
     ]
    }
   ],
   "source": [
    "train_model()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00026-2a696df7-60fd-4db7-bc88-5d920170a70e",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "> Now we need to do more feature engineering to try to improve our model. I have targeted a few columns that are not currently in use and could be added:\n",
    "\n",
    "    In previous_application:\n",
    "    * NAME_CASH_LOAN_PURPOSE\n",
    "    * NAME_CONTRACT_STATUS\n",
    "    * CODE_REJECT_REASON\n",
    "    * ... basically all of the potentially useful categorical features.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "cell_id": "00027-93eb6bbf-4b38-4302-9713-56240041ba45",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 153,
    "execution_start": 1616532239328,
    "source_hash": "54553979"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((307511, 530), 441212)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.shape, main_df.isnull().sum().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00028-50f11368-4b7c-4f16-9bc0-9f97827c9ffd",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "### Try again, using a different threshold for the categorical features encoding (method_switch=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "cell_id": "00029-d4b30a6b-7f42-4a9b-a772-9243d9a0f93d",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 269355,
    "execution_start": 1616532239473,
    "source_hash": "b084c0eb"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Memory usage of dataframe is 1.157779 GB\n",
      "Total initial memory used for selected dataframe is: 1.16GB\n",
      "Memory usage of dataframe is 1.107347 GB\n",
      "Total memory currently used for selected dataframe is: 1.11GB\n",
      "Memory usage of dataframe is 1.107347 GB\n",
      "Total initial memory used for selected dataframe is: 1.11GB\n",
      "Memory usage of dataframe is 0.590729 GB\n",
      "Total memory currently used for selected dataframe is: 0.59GB\n",
      "CPU times: user 10min 47s, sys: 1min 44s, total: 12min 32s\n",
      "Wall time: 13min 24s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "names = [re.sub('\\.csv$', '', ntpath.basename(p)) for p in glob.glob(data_dir + \"*.csv\")]\n",
    "\n",
    "dfs = {}\n",
    "for df_name in names:\n",
    "    dfs[df_name] = pd.read_csv(f\"{data_dir}{df_name}.csv\")\n",
    "\n",
    "for df in dfs:\n",
    "    dfs[df] = na_catfiller(dfs[df])\n",
    "    dfs[df] = str_one_hot_encoder(dfs[df], 10)\n",
    "    dfs[df] = na_numfiller(dfs[df])\n",
    "    \n",
    "dfs[\"application_train\"].shape\n",
    "\n",
    "# bureau_balance = dfs[\"bureau_balance\"].reset_index().groupby(['SK_ID_BUREAU', 'STATUS'])['MONTHS_BALANCE'].aggregate(lambda x: x.max()-x.min()).unstack()\n",
    "# bureau_balance.columns = [f\"{bureau_balance.columns.name}_{c}_MONTHS_BALANCE_MAX_MIN_DIFF\" for c in bureau_balance.columns]\n",
    "# bureau_balance.reset_index(level=0, inplace=True)\n",
    "\n",
    "dfs[\"bureau\"] = dfs[\"bureau_merged\"]\n",
    "for f in [f for f in dfs[\"bureau\"].columns if dfs[\"bureau\"][f].isnull().sum()>0]:\n",
    "    dfs[\"bureau\"][f] = dfs[\"bureau\"][f].fillna(dfs[\"bureau\"][f].median())\n",
    "\n",
    "main_df = dfs[\"application_train\"]\n",
    "for table in aggr_dicts:\n",
    "    main_df = merge_with_aggr(main_df, dfs[table], \"SK_ID_CURR\", aggr_dicts[table], table)\n",
    "\n",
    "na_numfiller(main_df, aggregation_func=\"median\")\n",
    "optimize_inttypes(main_df, specify=\"int32\")\n",
    "optimize_floattypes(main_df, specify=\"float32\")\n",
    "main_df = str_catencoder(main_df, method_switch=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "cell_id": "00030-edccaafb-7ea4-4d98-b4c2-29858aa8fef0",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 7,
    "execution_start": 1616532508830,
    "source_hash": "b1041925"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(307511, 530)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "cell_id": "00031-5b6ceb4a-1ba7-43c7-942f-982ed7974b4a",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 267566,
    "execution_start": 1616532508833,
    "source_hash": "7a162d9f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=600, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=600\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] num_threads is set=4, n_jobs=6 will be ignored. Current value: num_threads=4\n",
      "Training until validation scores don't improve for 300 rounds\n",
      "[200]\ttraining's auc: 0.772542\ttraining's binary_logloss: 0.27039\tvalid_1's auc: 0.751622\tvalid_1's binary_logloss: 0.274435\n",
      "[400]\ttraining's auc: 0.792048\ttraining's binary_logloss: 0.239388\tvalid_1's auc: 0.7626\tvalid_1's binary_logloss: 0.247919\n",
      "[600]\ttraining's auc: 0.811742\ttraining's binary_logloss: 0.230009\tvalid_1's auc: 0.77227\tvalid_1's binary_logloss: 0.243023\n",
      "[800]\ttraining's auc: 0.82729\ttraining's binary_logloss: 0.223514\tvalid_1's auc: 0.77802\tvalid_1's binary_logloss: 0.240857\n",
      "[1000]\ttraining's auc: 0.840604\ttraining's binary_logloss: 0.218136\tvalid_1's auc: 0.781403\tvalid_1's binary_logloss: 0.239658\n",
      "[1200]\ttraining's auc: 0.852471\ttraining's binary_logloss: 0.213396\tvalid_1's auc: 0.782834\tvalid_1's binary_logloss: 0.239081\n",
      "[1400]\ttraining's auc: 0.863382\ttraining's binary_logloss: 0.209006\tvalid_1's auc: 0.784008\tvalid_1's binary_logloss: 0.23867\n",
      "[1600]\ttraining's auc: 0.873251\ttraining's binary_logloss: 0.204893\tvalid_1's auc: 0.784451\tvalid_1's binary_logloss: 0.238502\n",
      "[1800]\ttraining's auc: 0.882453\ttraining's binary_logloss: 0.200949\tvalid_1's auc: 0.785222\tvalid_1's binary_logloss: 0.238253\n",
      "[2000]\ttraining's auc: 0.891078\ttraining's binary_logloss: 0.197161\tvalid_1's auc: 0.785417\tvalid_1's binary_logloss: 0.238189\n",
      "[2200]\ttraining's auc: 0.898842\ttraining's binary_logloss: 0.193536\tvalid_1's auc: 0.785441\tvalid_1's binary_logloss: 0.238131\n",
      "[2400]\ttraining's auc: 0.905986\ttraining's binary_logloss: 0.190048\tvalid_1's auc: 0.785643\tvalid_1's binary_logloss: 0.23804\n",
      "[2600]\ttraining's auc: 0.913196\ttraining's binary_logloss: 0.186566\tvalid_1's auc: 0.786192\tvalid_1's binary_logloss: 0.237874\n",
      "[2800]\ttraining's auc: 0.919493\ttraining's binary_logloss: 0.18324\tvalid_1's auc: 0.786207\tvalid_1's binary_logloss: 0.237874\n",
      "Early stopping, best iteration is:\n",
      "[2689]\ttraining's auc: 0.916154\ttraining's binary_logloss: 0.185061\tvalid_1's auc: 0.78624\tvalid_1's binary_logloss: 0.237854\n"
     ]
    }
   ],
   "source": [
    "train_model()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00032-1143cf3d-ede2-4138-9261-66694aea31f5",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "> This is actually worse than the previous version using `str_catencoder(main_df, method_switch=10)`. Let's try more feature engineering."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00033-108eb7e5-0782-4d60-90cf-99fd4e7e6d92",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "# KFold cross-validation & Feature Importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "cell_id": "00034-9268f883-ee72-41e5-9b2d-b192c59faff0",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 724209,
    "execution_start": 1616532776493,
    "source_hash": "52f80e2f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(276759, 528) (30752, 528) 276759 30752\n",
      "...... training 1th fold \n",
      "\n",
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=600, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=600\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] num_threads is set=4, n_jobs=4 will be ignored. Current value: num_threads=4\n",
      "Training until validation scores don't improve for 300 rounds\n",
      "[500]\ttraining's auc: 0.807745\ttraining's binary_logloss: 0.232599\tvalid_1's auc: 0.773322\tvalid_1's binary_logloss: 0.242672\n",
      "[1000]\ttraining's auc: 0.85027\ttraining's binary_logloss: 0.214658\tvalid_1's auc: 0.785082\tvalid_1's binary_logloss: 0.237683\n",
      "[1500]\ttraining's auc: 0.881641\ttraining's binary_logloss: 0.201572\tvalid_1's auc: 0.787594\tvalid_1's binary_logloss: 0.236627\n",
      "[2000]\ttraining's auc: 0.906354\ttraining's binary_logloss: 0.190189\tvalid_1's auc: 0.788482\tvalid_1's binary_logloss: 0.236323\n",
      "Early stopping, best iteration is:\n",
      "[2140]\ttraining's auc: 0.912608\ttraining's binary_logloss: 0.187153\tvalid_1's auc: 0.788657\tvalid_1's binary_logloss: 0.236249\n",
      "current best auc score is:0.7886573649811149\n",
      "...... training 2th fold \n",
      "\n",
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=600, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=600\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] num_threads is set=4, n_jobs=4 will be ignored. Current value: num_threads=4\n",
      "Training until validation scores don't improve for 300 rounds\n",
      "[500]\ttraining's auc: 0.809346\ttraining's binary_logloss: 0.232019\tvalid_1's auc: 0.766201\tvalid_1's binary_logloss: 0.24438\n",
      "[1000]\ttraining's auc: 0.851914\ttraining's binary_logloss: 0.213986\tvalid_1's auc: 0.777421\tvalid_1's binary_logloss: 0.239927\n",
      "[1500]\ttraining's auc: 0.882868\ttraining's binary_logloss: 0.201018\tvalid_1's auc: 0.779432\tvalid_1's binary_logloss: 0.239181\n",
      "Early stopping, best iteration is:\n",
      "[1676]\ttraining's auc: 0.892026\ttraining's binary_logloss: 0.196881\tvalid_1's auc: 0.779807\tvalid_1's binary_logloss: 0.239053\n",
      "current best auc score is:0.7798067152872417\n",
      "...... training 3th fold \n",
      "\n",
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=600, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=600\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] num_threads is set=4, n_jobs=4 will be ignored. Current value: num_threads=4\n",
      "Training until validation scores don't improve for 300 rounds\n",
      "[500]\ttraining's auc: 0.808545\ttraining's binary_logloss: 0.232435\tvalid_1's auc: 0.77306\tvalid_1's binary_logloss: 0.242431\n",
      "[1000]\ttraining's auc: 0.85146\ttraining's binary_logloss: 0.214453\tvalid_1's auc: 0.783541\tvalid_1's binary_logloss: 0.237907\n",
      "[1500]\ttraining's auc: 0.882666\ttraining's binary_logloss: 0.201293\tvalid_1's auc: 0.785535\tvalid_1's binary_logloss: 0.237094\n",
      "[2000]\ttraining's auc: 0.907379\ttraining's binary_logloss: 0.18995\tvalid_1's auc: 0.78584\tvalid_1's binary_logloss: 0.236981\n",
      "Early stopping, best iteration is:\n",
      "[1920]\ttraining's auc: 0.903863\ttraining's binary_logloss: 0.191654\tvalid_1's auc: 0.785939\tvalid_1's binary_logloss: 0.23695\n",
      "current best auc score is:0.7859391809997422\n",
      "...... training 4th fold \n",
      "\n",
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=600, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=600\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] num_threads is set=4, n_jobs=4 will be ignored. Current value: num_threads=4\n",
      "Training until validation scores don't improve for 300 rounds\n",
      "[500]\ttraining's auc: 0.808243\ttraining's binary_logloss: 0.232376\tvalid_1's auc: 0.771849\tvalid_1's binary_logloss: 0.243141\n",
      "[1000]\ttraining's auc: 0.850583\ttraining's binary_logloss: 0.214409\tvalid_1's auc: 0.782493\tvalid_1's binary_logloss: 0.23869\n",
      "[1500]\ttraining's auc: 0.882152\ttraining's binary_logloss: 0.201252\tvalid_1's auc: 0.784578\tvalid_1's binary_logloss: 0.237842\n",
      "[2000]\ttraining's auc: 0.90686\ttraining's binary_logloss: 0.189855\tvalid_1's auc: 0.784798\tvalid_1's binary_logloss: 0.237709\n",
      "[2500]\ttraining's auc: 0.926751\ttraining's binary_logloss: 0.179534\tvalid_1's auc: 0.785027\tvalid_1's binary_logloss: 0.237648\n",
      "Early stopping, best iteration is:\n",
      "[2231]\ttraining's auc: 0.916495\ttraining's binary_logloss: 0.18502\tvalid_1's auc: 0.785257\tvalid_1's binary_logloss: 0.237581\n",
      "current best auc score is:0.785256711742784\n",
      "...... training 5th fold \n",
      "\n",
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=600, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=600\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] num_threads is set=4, n_jobs=4 will be ignored. Current value: num_threads=4\n",
      "Training until validation scores don't improve for 300 rounds\n",
      "[500]\ttraining's auc: 0.807517\ttraining's binary_logloss: 0.2326\tvalid_1's auc: 0.77525\tvalid_1's binary_logloss: 0.242431\n",
      "[1000]\ttraining's auc: 0.850547\ttraining's binary_logloss: 0.214585\tvalid_1's auc: 0.787137\tvalid_1's binary_logloss: 0.237367\n",
      "[1500]\ttraining's auc: 0.881941\ttraining's binary_logloss: 0.201475\tvalid_1's auc: 0.789421\tvalid_1's binary_logloss: 0.236402\n",
      "Early stopping, best iteration is:\n",
      "[1698]\ttraining's auc: 0.89241\ttraining's binary_logloss: 0.196818\tvalid_1's auc: 0.789839\tvalid_1's binary_logloss: 0.236246\n",
      "current best auc score is:0.789838800829439\n"
     ]
    }
   ],
   "source": [
    "# get X and y, X input, y target: this is total data set X and y\n",
    "features = [f for f in main_df.columns if f not in ['SK_ID_CURR', 'TARGET']]\n",
    "X = main_df[features]\n",
    "y = main_df['TARGET']\n",
    "\n",
    "#split out for some test set as if it is future unseen data, using train_test_split for now\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, shuffle=True, random_state=10, test_size=0.1)\n",
    "print(X_train.shape,X_test.shape,len(y_train),len(y_test))\n",
    "\n",
    "# we reindex the objects\n",
    "X_train.reset_index(drop=True,inplace=True)\n",
    "X_test.reset_index(drop=True,inplace=True)\n",
    "y_train.reset_index(drop=True,inplace=True)\n",
    "y_test.reset_index(drop=True,inplace=True)\n",
    "\n",
    "#define your lightgbm model by initializing parameters\n",
    "model_lgb = lgb.LGBMClassifier(\n",
    "            n_jobs=4,\n",
    "            n_estimators=100000,\n",
    "            boost_from_average='false',\n",
    "            learning_rate=0.01,\n",
    "            num_leaves=64,\n",
    "            num_threads=4,\n",
    "            max_depth=-1,\n",
    "            tree_learner = \"serial\",\n",
    "            feature_fraction = 0.7,\n",
    "            bagging_freq = 5,\n",
    "            bagging_fraction = 0.7,\n",
    "            min_data_in_leaf=600,\n",
    "            silent=-1,\n",
    "            verbose=-1,\n",
    "            max_bin = 255,\n",
    "            bagging_seed = 11,\n",
    "            )\n",
    "\n",
    "#initialize KFold, we can use stratified KFold to keep the same imbalance ratio for target\n",
    "kf = StratifiedKFold(n_splits=5,shuffle=True,random_state=10)\n",
    "\n",
    "# Use KFold to train 5 models\n",
    "auc_scores = [] # save auc score for each fold\n",
    "models = [] #save model for each fold\n",
    "for i, (train_idx,valid_idx) in enumerate(kf.split(X_train,y_train)):\n",
    "    \n",
    "    print('...... training {}th fold \\n'.format(i+1))\n",
    "    tr_X = X_train.loc[train_idx]\n",
    "    tr_y = y_train.loc[train_idx]\n",
    "    \n",
    "    va_X = X_train.loc[valid_idx]\n",
    "    va_y = y_train.loc[valid_idx]\n",
    "    \n",
    "    model = model_lgb # you need to initialize your lgb model at each loop, otherwise it will overwrite\n",
    "    model.fit(tr_X, tr_y, eval_set=[(tr_X, tr_y), (va_X, va_y)], eval_metric = 'auc', verbose=500, early_stopping_rounds = 300)\n",
    "    \n",
    "    # calculate current auc after training the model\n",
    "    pred_va_y = model.predict_proba(va_X,num_iteration=model.best_iteration_)[:,1]\n",
    "    auc = roc_auc_score(va_y,pred_va_y)\n",
    "    print('current best auc score is:{}'.format(auc))\n",
    "    auc_scores.append(auc)\n",
    "    models.append(model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "cell_id": "00035-bb8b6885-5ec7-4eed-b655-edbe4bfe2e89",
    "deepnote_cell_type": "markdown"
   },
   "source": [
    "## Evaluate the models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "cell_id": "00036-1dfd6f3f-0d5a-4ad7-a276-9769dac59531",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 17853,
    "execution_start": 1616533500705,
    "source_hash": "a306aa03"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "the average mean auc is:0.7858997547680643\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([0.04882051, 0.07036165, 0.11738835, ..., 0.03381582, 0.02212431,\n",
       "       0.02424018])"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print('the average mean auc is:{}'.format(np.mean(auc_scores)))\n",
    "\n",
    "pred_test_1 = models[0].predict_proba(X_test,num_iteration=models[0].best_iteration_)[:,1]\n",
    "pred_test_2 = models[1].predict_proba(X_test,num_iteration=models[1].best_iteration_)[:,1]\n",
    "pred_test_3 = models[2].predict_proba(X_test,num_iteration=models[2].best_iteration_)[:,1]\n",
    "pred_test_4 = models[3].predict_proba(X_test,num_iteration=models[3].best_iteration_)[:,1]\n",
    "pred_test_5 = models[4].predict_proba(X_test,num_iteration=models[4].best_iteration_)[:,1]\n",
    "\n",
    "pred_test = (pred_test_1+pred_test_2+pred_test_3+pred_test_4+pred_test_5)/5.0\n",
    "\n",
    "models[0].predict(X_test,num_iteration=models[0].best_iteration_)\n",
    "\n",
    "models[0].predict_proba(X_test,num_iteration=models[0].best_iteration_)[:,1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "cell_id": "00037-e0f2744e-a200-4ab7-8bb4-46fb439a7660",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 7,
    "execution_start": 1616533518561,
    "source_hash": "ab17a10b"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['CNT_CHILDREN', 'AMT_INCOME_TOTAL', 'AMT_CREDIT', 'AMT_ANNUITY',\n",
       "       'AMT_GOODS_PRICE', 'REGION_POPULATION_RELATIVE', 'DAYS_BIRTH',\n",
       "       'DAYS_EMPLOYED', 'DAYS_REGISTRATION', 'DAYS_ID_PUBLISH',\n",
       "       'OWN_CAR_AGE', 'FLAG_MOBIL', 'FLAG_EMP_PHONE', 'FLAG_WORK_PHONE',\n",
       "       'FLAG_CONT_MOBILE', 'FLAG_PHONE', 'FLAG_EMAIL', 'CNT_FAM_MEMBERS',\n",
       "       'REGION_RATING_CLIENT', 'REGION_RATING_CLIENT_W_CITY',\n",
       "       'HOUR_APPR_PROCESS_START', 'REG_REGION_NOT_LIVE_REGION',\n",
       "       'REG_REGION_NOT_WORK_REGION', 'LIVE_REGION_NOT_WORK_REGION',\n",
       "       'REG_CITY_NOT_LIVE_CITY', 'REG_CITY_NOT_WORK_CITY',\n",
       "       'LIVE_CITY_NOT_WORK_CITY', 'EXT_SOURCE_1', 'EXT_SOURCE_2',\n",
       "       'EXT_SOURCE_3', 'APARTMENTS_AVG', 'BASEMENTAREA_AVG',\n",
       "       'YEARS_BEGINEXPLUATATION_AVG', 'YEARS_BUILD_AVG', 'COMMONAREA_AVG',\n",
       "       'ELEVATORS_AVG', 'ENTRANCES_AVG', 'FLOORSMAX_AVG', 'FLOORSMIN_AVG',\n",
       "       'LANDAREA_AVG', 'LIVINGAPARTMENTS_AVG', 'LIVINGAREA_AVG',\n",
       "       'NONLIVINGAPARTMENTS_AVG', 'NONLIVINGAREA_AVG', 'APARTMENTS_MODE',\n",
       "       'BASEMENTAREA_MODE', 'YEARS_BEGINEXPLUATATION_MODE',\n",
       "       'YEARS_BUILD_MODE', 'COMMONAREA_MODE', 'ELEVATORS_MODE',\n",
       "       'ENTRANCES_MODE', 'FLOORSMAX_MODE', 'FLOORSMIN_MODE',\n",
       "       'LANDAREA_MODE', 'LIVINGAPARTMENTS_MODE', 'LIVINGAREA_MODE',\n",
       "       'NONLIVINGAPARTMENTS_MODE', 'NONLIVINGAREA_MODE',\n",
       "       'APARTMENTS_MEDI', 'BASEMENTAREA_MEDI',\n",
       "       'YEARS_BEGINEXPLUATATION_MEDI', 'YEARS_BUILD_MEDI',\n",
       "       'COMMONAREA_MEDI', 'ELEVATORS_MEDI', 'ENTRANCES_MEDI',\n",
       "       'FLOORSMAX_MEDI', 'FLOORSMIN_MEDI', 'LANDAREA_MEDI',\n",
       "       'LIVINGAPARTMENTS_MEDI', 'LIVINGAREA_MEDI',\n",
       "       'NONLIVINGAPARTMENTS_MEDI', 'NONLIVINGAREA_MEDI', 'TOTALAREA_MODE',\n",
       "       'OBS_30_CNT_SOCIAL_CIRCLE', 'DEF_30_CNT_SOCIAL_CIRCLE',\n",
       "       'OBS_60_CNT_SOCIAL_CIRCLE', 'DEF_60_CNT_SOCIAL_CIRCLE',\n",
       "       'DAYS_LAST_PHONE_CHANGE', 'FLAG_DOCUMENT_2', 'FLAG_DOCUMENT_3',\n",
       "       'FLAG_DOCUMENT_4', 'FLAG_DOCUMENT_5', 'FLAG_DOCUMENT_6',\n",
       "       'FLAG_DOCUMENT_7', 'FLAG_DOCUMENT_8', 'FLAG_DOCUMENT_9',\n",
       "       'FLAG_DOCUMENT_10', 'FLAG_DOCUMENT_11', 'FLAG_DOCUMENT_12',\n",
       "       'FLAG_DOCUMENT_13', 'FLAG_DOCUMENT_14', 'FLAG_DOCUMENT_15',\n",
       "       'FLAG_DOCUMENT_16', 'FLAG_DOCUMENT_17', 'FLAG_DOCUMENT_18',\n",
       "       'FLAG_DOCUMENT_19', 'FLAG_DOCUMENT_20', 'FLAG_DOCUMENT_21',\n",
       "       'AMT_REQ_CREDIT_BUREAU_HOUR', 'AMT_REQ_CREDIT_BUREAU_DAY',\n",
       "       'AMT_REQ_CREDIT_BUREAU_WEEK', 'AMT_REQ_CREDIT_BUREAU_MON',\n",
       "       'AMT_REQ_CREDIT_BUREAU_QRT', 'AMT_REQ_CREDIT_BUREAU_YEAR',\n",
       "       'NAME_CONTRACT_TYPE_Cashloans',\n",
       "       'NAME_CONTRACT_TYPE_Revolvingloans', 'CODE_GENDER_F',\n",
       "       'CODE_GENDER_M', 'CODE_GENDER_XNA', 'FLAG_OWN_CAR_N',\n",
       "       'FLAG_OWN_CAR_Y', 'FLAG_OWN_REALTY_N', 'FLAG_OWN_REALTY_Y',\n",
       "       'NAME_TYPE_SUITE_Children', 'NAME_TYPE_SUITE_Family',\n",
       "       'NAME_TYPE_SUITE_Groupofpeople', 'NAME_TYPE_SUITE_Other_A',\n",
       "       'NAME_TYPE_SUITE_Other_B', 'NAME_TYPE_SUITE_Spousepartner',\n",
       "       'NAME_TYPE_SUITE_Unaccompanied', 'NAME_TYPE_SUITE_missing_data',\n",
       "       'NAME_INCOME_TYPE_Businessman',\n",
       "       'NAME_INCOME_TYPE_Commercialassociate',\n",
       "       'NAME_INCOME_TYPE_Maternityleave', 'NAME_INCOME_TYPE_Pensioner',\n",
       "       'NAME_INCOME_TYPE_Stateservant', 'NAME_INCOME_TYPE_Student',\n",
       "       'NAME_INCOME_TYPE_Unemployed', 'NAME_INCOME_TYPE_Working',\n",
       "       'NAME_EDUCATION_TYPE_Academicdegree',\n",
       "       'NAME_EDUCATION_TYPE_Highereducation',\n",
       "       'NAME_EDUCATION_TYPE_Incompletehigher',\n",
       "       'NAME_EDUCATION_TYPE_Lowersecondary',\n",
       "       'NAME_EDUCATION_TYPE_Secondarysecondaryspecial',\n",
       "       'NAME_FAMILY_STATUS_Civilmarriage', 'NAME_FAMILY_STATUS_Married',\n",
       "       'NAME_FAMILY_STATUS_Separated',\n",
       "       'NAME_FAMILY_STATUS_Singlenotmarried',\n",
       "       'NAME_FAMILY_STATUS_Unknown', 'NAME_FAMILY_STATUS_Widow',\n",
       "       'NAME_HOUSING_TYPE_Coopapartment',\n",
       "       'NAME_HOUSING_TYPE_Houseapartment',\n",
       "       'NAME_HOUSING_TYPE_Municipalapartment',\n",
       "       'NAME_HOUSING_TYPE_Officeapartment',\n",
       "       'NAME_HOUSING_TYPE_Rentedapartment',\n",
       "       'NAME_HOUSING_TYPE_Withparents',\n",
       "       'WEEKDAY_APPR_PROCESS_START_FRIDAY',\n",
       "       'WEEKDAY_APPR_PROCESS_START_MONDAY',\n",
       "       'WEEKDAY_APPR_PROCESS_START_SATURDAY',\n",
       "       'WEEKDAY_APPR_PROCESS_START_SUNDAY',\n",
       "       'WEEKDAY_APPR_PROCESS_START_THURSDAY',\n",
       "       'WEEKDAY_APPR_PROCESS_START_TUESDAY',\n",
       "       'WEEKDAY_APPR_PROCESS_START_WEDNESDAY',\n",
       "       'FONDKAPREMONT_MODE_missing_data',\n",
       "       'FONDKAPREMONT_MODE_notspecified',\n",
       "       'FONDKAPREMONT_MODE_orgspecaccount',\n",
       "       'FONDKAPREMONT_MODE_regoperaccount',\n",
       "       'FONDKAPREMONT_MODE_regoperspecaccount',\n",
       "       'HOUSETYPE_MODE_blockofflats', 'HOUSETYPE_MODE_missing_data',\n",
       "       'HOUSETYPE_MODE_specifichousing', 'HOUSETYPE_MODE_terracedhouse',\n",
       "       'WALLSMATERIAL_MODE_Block', 'WALLSMATERIAL_MODE_Mixed',\n",
       "       'WALLSMATERIAL_MODE_Monolithic', 'WALLSMATERIAL_MODE_Others',\n",
       "       'WALLSMATERIAL_MODE_Panel', 'WALLSMATERIAL_MODE_Stonebrick',\n",
       "       'WALLSMATERIAL_MODE_Wooden', 'WALLSMATERIAL_MODE_missing_data',\n",
       "       'EMERGENCYSTATE_MODE_No', 'EMERGENCYSTATE_MODE_Yes',\n",
       "       'EMERGENCYSTATE_MODE_missing_data',\n",
       "       'POS_CASH_balance_MONTHS_BALANCE_max',\n",
       "       'POS_CASH_balance_MONTHS_BALANCE_min',\n",
       "       'POS_CASH_balance_MONTHS_BALANCE_median',\n",
       "       'POS_CASH_balance_MONTHS_BALANCE_max_min_diff',\n",
       "       'POS_CASH_balance_CNT_INSTALMENT_max',\n",
       "       'POS_CASH_balance_CNT_INSTALMENT_min',\n",
       "       'POS_CASH_balance_CNT_INSTALMENT_median',\n",
       "       'POS_CASH_balance_CNT_INSTALMENT_max_min_diff',\n",
       "       'POS_CASH_balance_CNT_INSTALMENT_FUTURE_max',\n",
       "       'POS_CASH_balance_CNT_INSTALMENT_FUTURE_min',\n",
       "       'POS_CASH_balance_CNT_INSTALMENT_FUTURE_median',\n",
       "       'POS_CASH_balance_CNT_INSTALMENT_FUTURE_max_min_diff',\n",
       "       'POS_CASH_balance_SK_DPD_max', 'POS_CASH_balance_SK_DPD_min',\n",
       "       'POS_CASH_balance_SK_DPD_median',\n",
       "       'POS_CASH_balance_SK_DPD_max_min_diff',\n",
       "       'POS_CASH_balance_SK_DPD_DEF_max',\n",
       "       'POS_CASH_balance_SK_DPD_DEF_min',\n",
       "       'POS_CASH_balance_SK_DPD_DEF_median',\n",
       "       'POS_CASH_balance_SK_DPD_DEF_max_min_diff',\n",
       "       'POS_CASH_balance_NAME_CONTRACT_STATUS_Active_sum',\n",
       "       'POS_CASH_balance_NAME_CONTRACT_STATUS_Amortizeddebt_sum',\n",
       "       'POS_CASH_balance_NAME_CONTRACT_STATUS_Approved_sum',\n",
       "       'POS_CASH_balance_NAME_CONTRACT_STATUS_Completed_sum',\n",
       "       'POS_CASH_balance_NAME_CONTRACT_STATUS_Demand_sum',\n",
       "       'POS_CASH_balance_NAME_CONTRACT_STATUS_Returnedtothestore_sum',\n",
       "       'POS_CASH_balance_NAME_CONTRACT_STATUS_Signed_sum',\n",
       "       'credit_card_balance_MONTHS_BALANCE_max',\n",
       "       'credit_card_balance_MONTHS_BALANCE_min',\n",
       "       'credit_card_balance_MONTHS_BALANCE_median',\n",
       "       'credit_card_balance_MONTHS_BALANCE_max_min_diff',\n",
       "       'credit_card_balance_AMT_BALANCE_max',\n",
       "       'credit_card_balance_AMT_BALANCE_min',\n",
       "       'credit_card_balance_AMT_BALANCE_median',\n",
       "       'credit_card_balance_AMT_BALANCE_max_min_diff',\n",
       "       'credit_card_balance_AMT_CREDIT_LIMIT_ACTUAL_max',\n",
       "       'credit_card_balance_AMT_CREDIT_LIMIT_ACTUAL_min',\n",
       "       'credit_card_balance_AMT_CREDIT_LIMIT_ACTUAL_median',\n",
       "       'credit_card_balance_AMT_CREDIT_LIMIT_ACTUAL_max_min_diff',\n",
       "       'credit_card_balance_AMT_DRAWINGS_ATM_CURRENT_max',\n",
       "       'credit_card_balance_AMT_DRAWINGS_ATM_CURRENT_min',\n",
       "       'credit_card_balance_AMT_DRAWINGS_ATM_CURRENT_median',\n",
       "       'credit_card_balance_AMT_DRAWINGS_ATM_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_AMT_DRAWINGS_CURRENT_max',\n",
       "       'credit_card_balance_AMT_DRAWINGS_CURRENT_min',\n",
       "       'credit_card_balance_AMT_DRAWINGS_CURRENT_median',\n",
       "       'credit_card_balance_AMT_DRAWINGS_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_AMT_DRAWINGS_OTHER_CURRENT_max',\n",
       "       'credit_card_balance_AMT_DRAWINGS_OTHER_CURRENT_min',\n",
       "       'credit_card_balance_AMT_DRAWINGS_OTHER_CURRENT_median',\n",
       "       'credit_card_balance_AMT_DRAWINGS_OTHER_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_AMT_DRAWINGS_POS_CURRENT_max',\n",
       "       'credit_card_balance_AMT_DRAWINGS_POS_CURRENT_min',\n",
       "       'credit_card_balance_AMT_DRAWINGS_POS_CURRENT_median',\n",
       "       'credit_card_balance_AMT_DRAWINGS_POS_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_AMT_INST_MIN_REGULARITY_max',\n",
       "       'credit_card_balance_AMT_INST_MIN_REGULARITY_min',\n",
       "       'credit_card_balance_AMT_INST_MIN_REGULARITY_median',\n",
       "       'credit_card_balance_AMT_INST_MIN_REGULARITY_max_min_diff',\n",
       "       'credit_card_balance_AMT_PAYMENT_CURRENT_max',\n",
       "       'credit_card_balance_AMT_PAYMENT_CURRENT_min',\n",
       "       'credit_card_balance_AMT_PAYMENT_CURRENT_median',\n",
       "       'credit_card_balance_AMT_PAYMENT_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_AMT_PAYMENT_TOTAL_CURRENT_max',\n",
       "       'credit_card_balance_AMT_PAYMENT_TOTAL_CURRENT_min',\n",
       "       'credit_card_balance_AMT_PAYMENT_TOTAL_CURRENT_median',\n",
       "       'credit_card_balance_AMT_PAYMENT_TOTAL_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_AMT_RECEIVABLE_PRINCIPAL_max',\n",
       "       'credit_card_balance_AMT_RECEIVABLE_PRINCIPAL_min',\n",
       "       'credit_card_balance_AMT_RECEIVABLE_PRINCIPAL_median',\n",
       "       'credit_card_balance_AMT_RECEIVABLE_PRINCIPAL_max_min_diff',\n",
       "       'credit_card_balance_AMT_RECIVABLE_max',\n",
       "       'credit_card_balance_AMT_RECIVABLE_min',\n",
       "       'credit_card_balance_AMT_RECIVABLE_median',\n",
       "       'credit_card_balance_AMT_RECIVABLE_max_min_diff',\n",
       "       'credit_card_balance_AMT_TOTAL_RECEIVABLE_max',\n",
       "       'credit_card_balance_AMT_TOTAL_RECEIVABLE_min',\n",
       "       'credit_card_balance_AMT_TOTAL_RECEIVABLE_median',\n",
       "       'credit_card_balance_AMT_TOTAL_RECEIVABLE_max_min_diff',\n",
       "       'credit_card_balance_CNT_DRAWINGS_ATM_CURRENT_max',\n",
       "       'credit_card_balance_CNT_DRAWINGS_ATM_CURRENT_min',\n",
       "       'credit_card_balance_CNT_DRAWINGS_ATM_CURRENT_median',\n",
       "       'credit_card_balance_CNT_DRAWINGS_ATM_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_CNT_DRAWINGS_CURRENT_max',\n",
       "       'credit_card_balance_CNT_DRAWINGS_CURRENT_min',\n",
       "       'credit_card_balance_CNT_DRAWINGS_CURRENT_median',\n",
       "       'credit_card_balance_CNT_DRAWINGS_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_CNT_DRAWINGS_OTHER_CURRENT_max',\n",
       "       'credit_card_balance_CNT_DRAWINGS_OTHER_CURRENT_min',\n",
       "       'credit_card_balance_CNT_DRAWINGS_OTHER_CURRENT_median',\n",
       "       'credit_card_balance_CNT_DRAWINGS_OTHER_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_CNT_DRAWINGS_POS_CURRENT_max',\n",
       "       'credit_card_balance_CNT_DRAWINGS_POS_CURRENT_min',\n",
       "       'credit_card_balance_CNT_DRAWINGS_POS_CURRENT_median',\n",
       "       'credit_card_balance_CNT_DRAWINGS_POS_CURRENT_max_min_diff',\n",
       "       'credit_card_balance_CNT_INSTALMENT_MATURE_CUM_max',\n",
       "       'credit_card_balance_CNT_INSTALMENT_MATURE_CUM_min',\n",
       "       'credit_card_balance_CNT_INSTALMENT_MATURE_CUM_median',\n",
       "       'credit_card_balance_CNT_INSTALMENT_MATURE_CUM_max_min_diff',\n",
       "       'credit_card_balance_SK_DPD_max', 'credit_card_balance_SK_DPD_min',\n",
       "       'credit_card_balance_SK_DPD_median',\n",
       "       'credit_card_balance_SK_DPD_max_min_diff',\n",
       "       'credit_card_balance_SK_DPD_DEF_max',\n",
       "       'credit_card_balance_SK_DPD_DEF_min',\n",
       "       'credit_card_balance_SK_DPD_DEF_median',\n",
       "       'credit_card_balance_SK_DPD_DEF_max_min_diff',\n",
       "       'credit_card_balance_NAME_CONTRACT_STATUS_Active_sum',\n",
       "       'credit_card_balance_NAME_CONTRACT_STATUS_Completed_sum',\n",
       "       'credit_card_balance_NAME_CONTRACT_STATUS_Demand_sum',\n",
       "       'credit_card_balance_NAME_CONTRACT_STATUS_Signed_sum',\n",
       "       'installments_payments_NUM_INSTALMENT_VERSION_max',\n",
       "       'installments_payments_NUM_INSTALMENT_VERSION_min',\n",
       "       'installments_payments_NUM_INSTALMENT_VERSION_median',\n",
       "       'installments_payments_NUM_INSTALMENT_VERSION_max_min_diff',\n",
       "       'installments_payments_NUM_INSTALMENT_NUMBER_max',\n",
       "       'installments_payments_NUM_INSTALMENT_NUMBER_min',\n",
       "       'installments_payments_NUM_INSTALMENT_NUMBER_median',\n",
       "       'installments_payments_NUM_INSTALMENT_NUMBER_max_min_diff',\n",
       "       'installments_payments_DAYS_INSTALMENT_max',\n",
       "       'installments_payments_DAYS_INSTALMENT_min',\n",
       "       'installments_payments_DAYS_INSTALMENT_median',\n",
       "       'installments_payments_DAYS_INSTALMENT_max_min_diff',\n",
       "       'installments_payments_DAYS_ENTRY_PAYMENT_max',\n",
       "       'installments_payments_DAYS_ENTRY_PAYMENT_min',\n",
       "       'installments_payments_DAYS_ENTRY_PAYMENT_median',\n",
       "       'installments_payments_DAYS_ENTRY_PAYMENT_max_min_diff',\n",
       "       'installments_payments_AMT_INSTALMENT_max',\n",
       "       'installments_payments_AMT_INSTALMENT_min',\n",
       "       'installments_payments_AMT_INSTALMENT_median',\n",
       "       'installments_payments_AMT_INSTALMENT_max_min_diff',\n",
       "       'installments_payments_AMT_PAYMENT_max',\n",
       "       'installments_payments_AMT_PAYMENT_min',\n",
       "       'installments_payments_AMT_PAYMENT_median',\n",
       "       'installments_payments_AMT_PAYMENT_max_min_diff',\n",
       "       'previous_application_AMT_ANNUITY_max',\n",
       "       'previous_application_AMT_ANNUITY_min',\n",
       "       'previous_application_AMT_ANNUITY_median',\n",
       "       'previous_application_AMT_ANNUITY_max_min_diff',\n",
       "       'previous_application_AMT_APPLICATION_max',\n",
       "       'previous_application_AMT_APPLICATION_min',\n",
       "       'previous_application_AMT_APPLICATION_median',\n",
       "       'previous_application_AMT_APPLICATION_max_min_diff',\n",
       "       'previous_application_AMT_CREDIT_max',\n",
       "       'previous_application_AMT_CREDIT_min',\n",
       "       'previous_application_AMT_CREDIT_median',\n",
       "       'previous_application_AMT_CREDIT_max_min_diff',\n",
       "       'previous_application_AMT_DOWN_PAYMENT_max',\n",
       "       'previous_application_AMT_DOWN_PAYMENT_min',\n",
       "       'previous_application_AMT_DOWN_PAYMENT_median',\n",
       "       'previous_application_AMT_DOWN_PAYMENT_max_min_diff',\n",
       "       'previous_application_AMT_GOODS_PRICE_max',\n",
       "       'previous_application_AMT_GOODS_PRICE_min',\n",
       "       'previous_application_AMT_GOODS_PRICE_median',\n",
       "       'previous_application_AMT_GOODS_PRICE_max_min_diff',\n",
       "       'previous_application_HOUR_APPR_PROCESS_START_max',\n",
       "       'previous_application_HOUR_APPR_PROCESS_START_min',\n",
       "       'previous_application_HOUR_APPR_PROCESS_START_median',\n",
       "       'previous_application_HOUR_APPR_PROCESS_START_max_min_diff',\n",
       "       'previous_application_RATE_DOWN_PAYMENT_max',\n",
       "       'previous_application_RATE_DOWN_PAYMENT_min',\n",
       "       'previous_application_RATE_DOWN_PAYMENT_median',\n",
       "       'previous_application_RATE_DOWN_PAYMENT_max_min_diff',\n",
       "       'previous_application_RATE_INTEREST_PRIMARY_max',\n",
       "       'previous_application_RATE_INTEREST_PRIMARY_min',\n",
       "       'previous_application_RATE_INTEREST_PRIMARY_median',\n",
       "       'previous_application_RATE_INTEREST_PRIMARY_max_min_diff',\n",
       "       'previous_application_RATE_INTEREST_PRIVILEGED_max',\n",
       "       'previous_application_RATE_INTEREST_PRIVILEGED_min',\n",
       "       'previous_application_RATE_INTEREST_PRIVILEGED_median',\n",
       "       'previous_application_RATE_INTEREST_PRIVILEGED_max_min_diff',\n",
       "       'previous_application_DAYS_DECISION_max',\n",
       "       'previous_application_DAYS_DECISION_min',\n",
       "       'previous_application_DAYS_DECISION_median',\n",
       "       'previous_application_DAYS_DECISION_max_min_diff',\n",
       "       'previous_application_SELLERPLACE_AREA_max',\n",
       "       'previous_application_SELLERPLACE_AREA_min',\n",
       "       'previous_application_SELLERPLACE_AREA_median',\n",
       "       'previous_application_SELLERPLACE_AREA_max_min_diff',\n",
       "       'previous_application_CNT_PAYMENT_max',\n",
       "       'previous_application_CNT_PAYMENT_min',\n",
       "       'previous_application_CNT_PAYMENT_median',\n",
       "       'previous_application_CNT_PAYMENT_max_min_diff',\n",
       "       'previous_application_DAYS_FIRST_DRAWING_max',\n",
       "       'previous_application_DAYS_FIRST_DRAWING_min',\n",
       "       'previous_application_DAYS_FIRST_DRAWING_median',\n",
       "       'previous_application_DAYS_FIRST_DRAWING_max_min_diff',\n",
       "       'previous_application_DAYS_FIRST_DUE_max',\n",
       "       'previous_application_DAYS_FIRST_DUE_min',\n",
       "       'previous_application_DAYS_FIRST_DUE_median',\n",
       "       'previous_application_DAYS_FIRST_DUE_max_min_diff',\n",
       "       'previous_application_DAYS_LAST_DUE_1ST_VERSION_max',\n",
       "       'previous_application_DAYS_LAST_DUE_1ST_VERSION_min',\n",
       "       'previous_application_DAYS_LAST_DUE_1ST_VERSION_median',\n",
       "       'previous_application_DAYS_LAST_DUE_1ST_VERSION_max_min_diff',\n",
       "       'previous_application_DAYS_LAST_DUE_max',\n",
       "       'previous_application_DAYS_LAST_DUE_min',\n",
       "       'previous_application_DAYS_LAST_DUE_median',\n",
       "       'previous_application_DAYS_LAST_DUE_max_min_diff',\n",
       "       'previous_application_DAYS_TERMINATION_max',\n",
       "       'previous_application_DAYS_TERMINATION_min',\n",
       "       'previous_application_DAYS_TERMINATION_median',\n",
       "       'previous_application_DAYS_TERMINATION_max_min_diff',\n",
       "       'previous_application_NAME_CONTRACT_TYPE_Cashloans_sum',\n",
       "       'previous_application_NAME_CONTRACT_TYPE_Consumerloans_sum',\n",
       "       'previous_application_NAME_CONTRACT_TYPE_Revolvingloans_sum',\n",
       "       'previous_application_NAME_CONTRACT_TYPE_XNA_sum',\n",
       "       'previous_application_WEEKDAY_APPR_PROCESS_START_FRIDAY_sum',\n",
       "       'previous_application_WEEKDAY_APPR_PROCESS_START_MONDAY_sum',\n",
       "       'previous_application_WEEKDAY_APPR_PROCESS_START_SATURDAY_sum',\n",
       "       'previous_application_WEEKDAY_APPR_PROCESS_START_SUNDAY_sum',\n",
       "       'previous_application_WEEKDAY_APPR_PROCESS_START_THURSDAY_sum',\n",
       "       'previous_application_WEEKDAY_APPR_PROCESS_START_TUESDAY_sum',\n",
       "       'previous_application_WEEKDAY_APPR_PROCESS_START_WEDNESDAY_sum',\n",
       "       'previous_application_FLAG_LAST_APPL_PER_CONTRACT_N_sum',\n",
       "       'previous_application_FLAG_LAST_APPL_PER_CONTRACT_Y_sum',\n",
       "       'previous_application_NAME_CONTRACT_STATUS_Approved_sum',\n",
       "       'previous_application_NAME_CONTRACT_STATUS_Canceled_sum',\n",
       "       'previous_application_NAME_CONTRACT_STATUS_Refused_sum',\n",
       "       'previous_application_NAME_CONTRACT_STATUS_Unusedoffer_sum',\n",
       "       'previous_application_NAME_PAYMENT_TYPE_Cashthroughthebank_sum',\n",
       "       'previous_application_NAME_PAYMENT_TYPE_Cashlessfromtheaccountoftheemployer_sum',\n",
       "       'previous_application_NAME_PAYMENT_TYPE_Noncashfromyouraccount_sum',\n",
       "       'previous_application_NAME_PAYMENT_TYPE_XNA_sum',\n",
       "       'previous_application_CODE_REJECT_REASON_CLIENT_sum',\n",
       "       'previous_application_CODE_REJECT_REASON_HC_sum',\n",
       "       'previous_application_CODE_REJECT_REASON_LIMIT_sum',\n",
       "       'previous_application_CODE_REJECT_REASON_SCO_sum',\n",
       "       'previous_application_CODE_REJECT_REASON_SCOFR_sum',\n",
       "       'previous_application_CODE_REJECT_REASON_SYSTEM_sum',\n",
       "       'previous_application_CODE_REJECT_REASON_VERIF_sum',\n",
       "       'previous_application_CODE_REJECT_REASON_XAP_sum',\n",
       "       'previous_application_CODE_REJECT_REASON_XNA_sum',\n",
       "       'previous_application_NAME_TYPE_SUITE_Children_sum',\n",
       "       'previous_application_NAME_TYPE_SUITE_Family_sum',\n",
       "       'previous_application_NAME_TYPE_SUITE_Groupofpeople_sum',\n",
       "       'previous_application_NAME_TYPE_SUITE_Other_A_sum',\n",
       "       'previous_application_NAME_TYPE_SUITE_Other_B_sum',\n",
       "       'previous_application_NAME_TYPE_SUITE_Spousepartner_sum',\n",
       "       'previous_application_NAME_TYPE_SUITE_Unaccompanied_sum',\n",
       "       'previous_application_NAME_CLIENT_TYPE_New_sum',\n",
       "       'previous_application_NAME_CLIENT_TYPE_Refreshed_sum',\n",
       "       'previous_application_NAME_CLIENT_TYPE_Repeater_sum',\n",
       "       'previous_application_NAME_CLIENT_TYPE_XNA_sum',\n",
       "       'previous_application_NAME_PORTFOLIO_Cards_sum',\n",
       "       'previous_application_NAME_PORTFOLIO_Cars_sum',\n",
       "       'previous_application_NAME_PORTFOLIO_Cash_sum',\n",
       "       'previous_application_NAME_PORTFOLIO_POS_sum',\n",
       "       'previous_application_NAME_PORTFOLIO_XNA_sum',\n",
       "       'previous_application_NAME_PRODUCT_TYPE_XNA_sum',\n",
       "       'previous_application_NAME_PRODUCT_TYPE_walkin_sum',\n",
       "       'previous_application_NAME_PRODUCT_TYPE_xsell_sum',\n",
       "       'previous_application_CHANNEL_TYPE_APCashloan_sum',\n",
       "       'previous_application_CHANNEL_TYPE_Cardealer_sum',\n",
       "       'previous_application_CHANNEL_TYPE_Channelofcorporatesales_sum',\n",
       "       'previous_application_CHANNEL_TYPE_Contactcenter_sum',\n",
       "       'previous_application_CHANNEL_TYPE_Countrywide_sum',\n",
       "       'previous_application_CHANNEL_TYPE_Creditandcashoffices_sum',\n",
       "       'previous_application_CHANNEL_TYPE_RegionalLocal_sum',\n",
       "       'previous_application_CHANNEL_TYPE_Stone_sum',\n",
       "       'previous_application_NAME_YIELD_GROUP_XNA_sum',\n",
       "       'previous_application_NAME_YIELD_GROUP_high_sum',\n",
       "       'previous_application_NAME_YIELD_GROUP_low_action_sum',\n",
       "       'previous_application_NAME_YIELD_GROUP_low_normal_sum',\n",
       "       'previous_application_NAME_YIELD_GROUP_middle_sum',\n",
       "       'bureau_SK_ID_BUREAU_max', 'bureau_SK_ID_BUREAU_min',\n",
       "       'bureau_SK_ID_BUREAU_median', 'bureau_SK_ID_BUREAU_max_min_diff',\n",
       "       'bureau_DAYS_CREDIT_max', 'bureau_DAYS_CREDIT_min',\n",
       "       'bureau_DAYS_CREDIT_median', 'bureau_DAYS_CREDIT_max_min_diff',\n",
       "       'bureau_CREDIT_DAY_OVERDUE_max', 'bureau_CREDIT_DAY_OVERDUE_min',\n",
       "       'bureau_CREDIT_DAY_OVERDUE_median',\n",
       "       'bureau_CREDIT_DAY_OVERDUE_max_min_diff',\n",
       "       'bureau_DAYS_CREDIT_ENDDATE_max', 'bureau_DAYS_CREDIT_ENDDATE_min',\n",
       "       'bureau_DAYS_CREDIT_ENDDATE_median',\n",
       "       'bureau_DAYS_CREDIT_ENDDATE_max_min_diff',\n",
       "       'bureau_DAYS_ENDDATE_FACT_max', 'bureau_DAYS_ENDDATE_FACT_min',\n",
       "       'bureau_DAYS_ENDDATE_FACT_median',\n",
       "       'bureau_DAYS_ENDDATE_FACT_max_min_diff',\n",
       "       'bureau_AMT_CREDIT_MAX_OVERDUE_max',\n",
       "       'bureau_AMT_CREDIT_MAX_OVERDUE_min',\n",
       "       'bureau_AMT_CREDIT_MAX_OVERDUE_median',\n",
       "       'bureau_AMT_CREDIT_MAX_OVERDUE_max_min_diff',\n",
       "       'bureau_CNT_CREDIT_PROLONG_max', 'bureau_CNT_CREDIT_PROLONG_min',\n",
       "       'bureau_CNT_CREDIT_PROLONG_median',\n",
       "       'bureau_CNT_CREDIT_PROLONG_max_min_diff',\n",
       "       'bureau_AMT_CREDIT_SUM_max', 'bureau_AMT_CREDIT_SUM_min',\n",
       "       'bureau_AMT_CREDIT_SUM_median',\n",
       "       'bureau_AMT_CREDIT_SUM_max_min_diff',\n",
       "       'bureau_AMT_CREDIT_SUM_DEBT_max', 'bureau_AMT_CREDIT_SUM_DEBT_min',\n",
       "       'bureau_AMT_CREDIT_SUM_DEBT_median',\n",
       "       'bureau_AMT_CREDIT_SUM_DEBT_max_min_diff',\n",
       "       'bureau_AMT_CREDIT_SUM_LIMIT_max',\n",
       "       'bureau_AMT_CREDIT_SUM_LIMIT_min',\n",
       "       'bureau_AMT_CREDIT_SUM_LIMIT_median',\n",
       "       'bureau_AMT_CREDIT_SUM_LIMIT_max_min_diff',\n",
       "       'bureau_AMT_CREDIT_SUM_OVERDUE_max',\n",
       "       'bureau_AMT_CREDIT_SUM_OVERDUE_min',\n",
       "       'bureau_AMT_CREDIT_SUM_OVERDUE_median',\n",
       "       'bureau_AMT_CREDIT_SUM_OVERDUE_max_min_diff',\n",
       "       'bureau_DAYS_CREDIT_UPDATE_max', 'bureau_DAYS_CREDIT_UPDATE_min',\n",
       "       'bureau_DAYS_CREDIT_UPDATE_median',\n",
       "       'bureau_DAYS_CREDIT_UPDATE_max_min_diff', 'bureau_AMT_ANNUITY_max',\n",
       "       'bureau_AMT_ANNUITY_min', 'bureau_AMT_ANNUITY_median',\n",
       "       'bureau_AMT_ANNUITY_max_min_diff',\n",
       "       'bureau_STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "       'bureau_STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "       'bureau_STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "       'bureau_STATUS_0_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "       'bureau_STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "       'bureau_STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "       'bureau_STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "       'bureau_STATUS_1_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "       'bureau_STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "       'bureau_STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "       'bureau_STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "       'bureau_STATUS_2_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "       'bureau_STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "       'bureau_STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "       'bureau_STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "       'bureau_STATUS_3_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "       'bureau_STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "       'bureau_STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "       'bureau_STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "       'bureau_STATUS_4_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "       'bureau_STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "       'bureau_STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "       'bureau_STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "       'bureau_STATUS_5_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "       'bureau_STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "       'bureau_STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "       'bureau_STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "       'bureau_STATUS_C_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "       'bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max',\n",
       "       'bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_min',\n",
       "       'bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_median',\n",
       "       'bureau_STATUS_X_MONTHS_BALANCE_MAX_MIN_DIFF_max_min_diff',\n",
       "       'bureau_CREDIT_ACTIVE_Active_sum',\n",
       "       'bureau_CREDIT_ACTIVE_Closed_sum', 'bureau_CREDIT_ACTIVE_Sold_sum',\n",
       "       'bureau_CREDIT_CURRENCY_currency1_sum',\n",
       "       'bureau_CREDIT_CURRENCY_currency2_sum',\n",
       "       'bureau_CREDIT_CURRENCY_currency3_sum'], dtype=object)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fts = X_train.columns.values # all features currently used\n",
    "fts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "cell_id": "00038-1324c016-3da4-4dce-95c5-a2d56d64ed01",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 10,
    "execution_start": 1616533518566,
    "source_hash": "1209fc8c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('EXT_SOURCE_2', 2771),\n",
       " ('EXT_SOURCE_1', 2357),\n",
       " ('EXT_SOURCE_3', 2259),\n",
       " ('DAYS_BIRTH', 2179),\n",
       " ('AMT_ANNUITY', 1917),\n",
       " ('AMT_CREDIT', 1792),\n",
       " ('installments_payments_AMT_PAYMENT_min', 1785),\n",
       " ('DAYS_EMPLOYED', 1492),\n",
       " ('AMT_GOODS_PRICE', 1464),\n",
       " ('DAYS_ID_PUBLISH', 1366),\n",
       " ('bureau_DAYS_CREDIT_max', 1232),\n",
       " ('DAYS_REGISTRATION', 1185),\n",
       " ('bureau_DAYS_CREDIT_ENDDATE_median', 1091),\n",
       " ('bureau_DAYS_CREDIT_ENDDATE_max', 1054),\n",
       " ('DAYS_LAST_PHONE_CHANGE', 1030),\n",
       " ('installments_payments_AMT_PAYMENT_median', 998),\n",
       " ('installments_payments_DAYS_ENTRY_PAYMENT_max', 949),\n",
       " ('bureau_AMT_CREDIT_SUM_median', 876),\n",
       " ('POS_CASH_balance_CNT_INSTALMENT_FUTURE_median', 862),\n",
       " ('bureau_DAYS_CREDIT_median', 833)]"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fts_imp = dict(zip(fts,models[2].feature_importances_))\n",
    "fts_imp = sorted(fts_imp.items(), key=operator.itemgetter(1),reverse=True)\n",
    "fts_imp[:20]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "cell_id": "00039-97838d4d-6a21-4c29-9d0f-5baadc345dab",
    "deepnote_cell_type": "code",
    "deepnote_to_be_reexecuted": false,
    "execution_millis": 569,
    "execution_start": 1616533518573,
    "source_hash": "debc138b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#print feature importance\n",
    "feature_imp = pd.DataFrame({'Value':models[0].feature_importances_,'Feature':X_train.columns})\n",
    "plt.figure(figsize=(20, 10))\n",
    "sns.set(font_scale = 2)\n",
    "sns.barplot(x=\"Value\", y=\"Feature\", data=feature_imp.sort_values(by=\"Value\", \n",
    "                                                    ascending=False)[:20])\n",
    "plt.title('LightGBM Features (avg over folds)')\n",
    "plt.tight_layout()\n",
    "#plt.savefig('lgbm_importances-01.png')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "deepnote": {},
  "deepnote_execution_queue": [],
  "deepnote_notebook_id": "1e8b86d6-17bd-4995-a8f0-97a801719f7f",
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}